The rapid expansion of wind power development in recent years has accentuated the need to develop standard guidelines for identifying, assessing, and monitoring potential impacts to birds and bats. Although postconstruction mortality estimates generally take into account well-established sources of bias, including searcher efficiency and scavenging loss, methods for addressing these biases can be improved. Currently used bias-adjustment methods differ across studies, do not explicitly account for factors that may affect initial bias estimates, and often use averaged or assumed levels of bias. We examined scavenging and detection trial data from a 3-y study at a small, terrestrial wind-farm in coastal New Jersey. Logistic regression models indicated that carcass size, substrate, and observer all affected carcass detection rates, with larger carcasses more likely to be detected than smaller carcasses, and those located on bare ground or grass more likely to be detected than those on gravel. Known-fate mark–recapture models indicated that scavenging rates were highest within the first 3 d of placement, with some variation among seasons. We suggest that empirically based estimates of factors affecting observer detection and scavenging loss be generated for individual wind-farm mortality studies, because they likely vary across sites and could heavily bias resulting adjustment factors and mortality estimates.
As the demand for renewable sources of energy continues to increase in the United States, so too will the need to understand how these rapidly growing sectors impact wildlife populations. Wind power has been used to commercially produce energy in the United States since the early 1980s and has grown exponentially as an industry. By the end of 2013, there were >900 utility-scale wind facilities operating in 39 U.S. states and Puerto Rico (AWEA 2014). The average height and size of wind turbines have also increased over time (ABC 2011; Wiser and Bolinger 2012). These developments have led to concern about potential negative impacts of wind power development on wildlife, particularly migratory birds and bats, and have prompted the development of standard guidelines for identifying, assessing, and monitoring those potential impacts (USFWS 2012).
As with any large structures on the landscape, wind turbines can be hazardous to organisms that use the airspace around them (review in Kuvlesky et al. 2007). Bat mortality, for example, has been documented in several postconstruction studies in the United States (Johnson et al. 2004; Arnett et al. 2008; Piorkowski and O'Connell 2010) and Europe (Rydell et al. 2010), mainly caused by collision with blades (Johnson et al. 2004; Baerwald et al. 2008; Cryan and Barclay 2009). Raptors are also susceptible to injury or death by wind-turbines (Hunt 2002; Hoover and Morrison 2005; Smallwood and Thelander 2008), as are migratory songbirds and shorebirds, though less is known about the extent of mortality in these groups (Osborn et al. 2000; Johnson et al. 2002; Kerlinger et al. 2010).
The Wind Turbine Guidelines Advisory Committee established by the U.S. Fish and Wildlife Service (USFWS) in 2007 developed a multitiered approach for assessing the impacts of wind power development on wildlife, from siting evaluations to postconstruction mortality verification (USFWS 2012). Postconstruction mortality estimates generally take into account several well-established sources of bias, including searcher efficiency and scavenging loss (Kunz et al. 2007; Arnett et al. 2008). The extent of incomplete detection and scavenging bias is typically determined through the implementation of trial studies, in which the fates of randomly placed carcasses are monitored and used to estimate the proportion of carcasses that are detected, and the average number of days carcasses remain in the study area before being removed by scavengers (Kunz et al. 2007; Smallwood 2007; Arnett et al. 2008).
Studies implementing bias-adjustment methods sometimes acknowledge, although typically do not directly evaluate, disparity in observer efficiencies (Morrison et al. 2001) or the effects of factors such as habitat, season, or carcass type on scavenging and detection rates (Morrison 2002; Johnson et al. 2003; Kerns et al. 2005; Erickson 2006; Arnett et al. 2008; but see Smallwood et al. 2010; Bispo et al. 2013). Various methods to account for these potential biases have been employed, including averaging searcher efficiency rates (e.g., by observer and substrate) for incorporation into mortality estimation equations (see reviews in Arnett et al. 2008; Smallwood 2007), or generating detection or scavenging estimates separately for different carcass size classes and seasons (Smallwood et al. 2010). These methods are often applied subjectively, adjusting for some assumed biases while ignoring others without explicitly testing which are actually significant in a given study (reviews in Smallwood 2007; Bispo et al. 2013).
Binomial models represent one suite of methods that can be used to directly assess the relationships among independent factors and searcher efficiency and scavenging rates. For instance, in carcass detection trials, logistic regression methods (Hosmer and Lemeshow 2000) can be used to evaluate factors related to detection probability by expressing detection as a linear-logistic function of one or more independent variables (Stevens et al. 2011). Capture–recapture models, long used for estimating survival rates in animal populations (White and Burnham 1999) can similarly be used to estimate daily scavenging probabilities (i.e., 1 − carcass “survival” [ϕ] or persistence, Bispo et al. 2013; Peron et al. 2013). When encounter probabilities are assumed to be 1.0 (i.e., the status of each individual is known with certainty, as is the case in scavenging trials), known-fate models offer particular advantages in that precision of parameter estimates is high, even with small sample sizes (Cooch and White 2011). These methods also provide a basis for model comparison and model averaging in an information-theoretic context (Burnham and Anderson 2002), so that factors most strongly associated with searcher efficiency and scavenging rates can be identified (Bispo et al. 2013).
We conducted efficiency and scavenging trials during a 3-y study at a small, terrestrial wind-farm in coastal New Jersey. Focusing on several factors considered likely to influence detection and scavenge rate estimates, we used logistic regression (detection) and known-fate mark–recapture (scavenging) methods to explicitly model the magnitude and direction of these effects. Specific objectives were to 1) determine whether and to what extent carcass size, observer, and substrate affected detection probability estimates; and 2) determine whether carcass age, carcass size, landscape context, season, and carcass taxon affected scavenging rates.
The study was conducted at the Jersey Atlantic Wind, LLC/Atlantic City Utilities Authority (ACUA) wind power facility, near Atlantic City, New Jersey (39°22′N, 74°27′W). The facility serves as a wastewater treatment plant powered by five wind turbines that have been operational since December 2005. Each tower is 80 m tall with a diameter of 4 m. The blades are 37 m long, and the tips can reach an estimated 289 km/h at maximum wind speeds (ACUA 2014). Red navigation lights are located at the top of each turbine, as mandated by the Federal Aviation Administration. The facility is juxtaposed within a tidal marsh dominated by cordgrass (Spartina alterniflora, S. patens), and common reed Phragmites australis. The Edwin B. Forsythe National Wildlife Refuge is located ∼6 km north of the wind facility, and consists of >19,000 ha of coastal habitats that are actively managed for migratory birds. The site lies within the Atlantic Flyway along the southern coast of New Jersey, a key migratory corridor for raptors (Farmer et al. 2007), songbirds (Wiedner et al. 1992), shorebirds (Clark et al. 1993), seabirds (Mizrahi 2009), and migratory tree bats Lasiurus spp. (Cryan 2003).
Search plots centered on each of the five turbines were established prior to the efficiency and scavenging trials. Each search plot consisted of a 130-m-per-side square to ensure that all area within 63 m of the turbine was included within the plot boundary (84,500 m2 [Johnson et al. 2000; Jain et al. 2009]), and was oriented to provide maximum searchable area within the landscape context. We set the 63-m distance conservatively, because avian fatalities have rarely been documented up to 63 m from turbines at other wind plants (Johnson et al. 2000) and bat fatalities tend to occur within 30 m of turbines (Johnson et al. 2004). The searcher efficiency and scavenging trials were conducted within the context of a postconstruction mortality study (Peron et al. 2013), during which an observer searched each plot for bird and bat carcasses on an approximate 3-d/wk schedule from August 2007 through July 2010. On each search day, one observer traversed transects established within plots. Transects were spaced 5 m apart, ran parallel to a plot side, and were marked at the ends prior to the field study. Searcher efficiency and scavenging trial dates were not disclosed beforehand to the observers.
We conducted efficiency and scavenging trials from September 2007 through June 2010, with a maximum of one trial conducted per month. In each trial, we randomly placed 3–5 carcasses (X¯ = 4.00, SD = 0.46) within the searchable area of the study site and within 24 h prior to a scheduled carcass search. We placed a maximum of two carcasses per turbine to avoid “overseeding” an area and potentially biasing scavenging rates low (Smallwood et al. 2010; USFWS 2012). Most carcasses had been frozen prior to the trial period, and represented intact turbine strikes from previous searches or birds turned in to the Cape May Bird Observatory by the public. A varied distribution of avian and bat sizes was represented in each trial (29 bats, 55 birds total). Observers were notified of the trial at the end of the corresponding day's carcass search, at which time they were given coordinates to each undetected carcass to determine whether it had been scavenged or remained on site; if the carcass was not located immediately, the observer walked concentric circles around the initial placement point to ensure that it had been removed. Scavenged carcasses were assumed to have been removed prior to that day's search. However, we acknowledge that if scavenging occurred during the 4–5-h search period (i.e., after a turbine had been searched, but before carcass presence or absence was confirmed), searcher efficiency could have been overestimated.
Each carcass was recorded according to whether it had been detected by the observer during the initial search (1 = detected, 0 = not detected), and whether it had been scavenged (1 = scavenged, 0 = not scavenged). Carcasses assumed to have been scavenged prior to the search were not included in estimates of searcher efficiency. All other carcasses were checked daily for 1 wk, from which we created encounter histories for each marked individual for use in survival analysis (i.e., scavenge-rate estimation). We selected a short interval period between search days (i.e., 1 d) in order to account for small bird and bat removals, which typically occur on a smaller time scale (i.e., more rapidly) than removals of large birds such as hawks and eagles (USFWS 2012). Twenty trial periods were completed, involving the placement of 80 carcasses (autumn [September–November], 6 trials; winter [December–March], 8 trials; spring [April–June], 6 trials; summer [July–August], 6 trials).
We used logistic regression models to estimate detection probability and quantify factors that affected detection estimates (Proc Logistic, SAS v 9.2; SAS Institute, Inc.). We used data from 59 carcasses in the analysis (Table S1, Supplemental Material). We examined all possible independent (i.e., nonnested) candidate models with ≤3 explanatory variables (due to our limited sample size, n = 25 detections, or “events”). “Detected” (D = 1) or “Not Detected” (D = 0) served as the response variable, and explanatory variables included “Observer” (n = 3), “Substrate” (grass, gravel, bare ground), “Mass” (mean carcass species mass), and “Temperature” (mean daily temperature; Atlantic City International Airport, National Oceanic and Atmospheric Administration). Carcass mass estimates were based on mean species-specific measurements reported in the literature (Table 1). Substrate was initially determined using a Geographic Information System (GIS, ArcView 9.3; ESRI, Inc.) and ground-truthed after placement. We used an information-theoretic approach to determine the best model(s) for predicting detectability (Burnham and Anderson 2002). Models that were within 3 corrected Akaike Information Criterion (AICc) points of the best performing model were considered competitive. Parameter estimates were averaged according to model weights, model-averaged variance was computed using the estimator described in Burnham and Anderson (2002; Equation 4.11), and all tests for differences among parameter levels were 2-tailed.
We conducted scavenging rate analyses on 80 trial carcasses (Table S2, Supplemental Material). We used the binomial known-fate model module within Program MARK (v. 6.2; White and Burnham 1999) to estimate carcass persistence rates and identify parameters that influenced removal rates. Individual parameter variances were estimated in Program MARK (White and Burnham 1999), and variance associated with cumulative persistence was estimated using the delta method. We examined nine models that estimated carcass daily persistence rates, or the likelihood that a carcass was not scavenged between daily searches. We assumed a priori that persistence probability would be affected by the amount of time that had passed since a carcass had been placed (Smallwood et al. 2010; Bispo et al. 2013; Peron et al. 2013) and examined all one- and two-predictor variable models that included the variable “Age” (days since placement). Candidate models incorporated the variables “Season”, “Distance” (i.e., distance from nearest marsh edge, calculated in a GIS), “Taxa” (i.e., bird or bat), “Temperature” (mean daily temperature; Atlantic City International Airport, National Oceanic and Atmospheric Administration), “Mass” (mean mass of carcass species; Table 1), and a “Season” × “Age” interaction. A null, or base model, included only the intercept. No models with more than two predictor variables, including a global model, were constructed because of sample size limitations. Parameter estimates were model-averaged when appropriate and presented as such, all confidence intervals reported represent 95% Confidence Interval, and significance was accepted when P ≤ 0.05.
Searcher efficiency trials
Of the 80 carcasses initially placed, 59 remained (i.e., were not scavenged) on day one and we subsequently included them in searcher efficiency analyses. Two models that collectively included parameters for mass, substrate, and observer performed best for predicting whether a carcass was detected during search trials (Table 2). Based on these models, as carcasses increased in size they were significantly more likely to be detected (Table 3). Carcasses placed on bare ground were more likely to be found than those placed on gravel, but not more likely to be detected than those placed on grass (Table 3). Detection rates did not differ between grass and gravel substrates (Wald-χ2 = 0.08, P = 0.78). The model also identified a significant difference in observer detection rates, with “Searcher 1” less likely to detect carcasses than “Searcher 3” (Table 3), and somewhat less likely than “Searcher 2” (P = 0.08).
The best-performing models for predicting likelihood of carcass daily persistence all included the parameters “Age” (i.e., days since placement; Table 4). Further examination of the data indicated that daily persistence rate was time-dependent (time since death; in this study carcass placement); model-averaged daily persistence estimates for all seasons combined increased from 73.1% to nearly 100% over the search period (Figure 1a). Based on cumulative persistence calculations, an average of 39.5% of carcasses killed on site would be predicted to remain after 7 d, although this estimate was imprecise (Figure 1b; CI = 28.5–50.5%). Under the assumption that daily persistence remained relatively constant after day 7 (daily persistence rate = 0.968) we extrapolated cumulative persistence rate to 14 d, with an estimated result of 31.4% (CI = 21.5–41.2%) of carcasses predicted to remain on site (Figure 1b). This assumption was based on the relatively stable daily persistence rates observed between days 5 and 7; however, it does not preclude the possibility that daily persistence rates began to decline again after day 6, as also suggested by the data (potentially due to carcass decomposition or other processes that would make large carcasses more accessible to scavengers). Several parameters at the boundaries of the “Age × Season” model were inestimable (i.e., only 13 of 28 parameters were estimable) due to data limitations, rendering the model uncompetitive. However, parameter estimates from the top performing models indicated that an overall “Season” trend may have been present, with scavenging rates lowest during winter months (Table 5).
By using binomial modeling techniques, we were able to directly estimate the effects of various factors on searcher efficiency and scavenging rates. A previous analysis of the ACUA data indicated that neglecting the finite persistence time and imperfect detection probability of carcasses could lead to an approximate two-fold underestimation of fatality number if not properly modeled in the estimation framework (Peron et al. 2013). Therefore, our aim was to provide a comprehensive, quantitative case study illustrating several sources of variation that could bias detection, persistence, and resultant fatality estimates. Model results clearly demonstrated that larger carcasses were more detectable than small carcasses. Prior studies have determined that larger carcasses are more likely to be seen by observers (Osborn et al. 2000; Smallwood 2007), and we were able to explicitly model this process, provide parameter estimates (and associated confidence intervals), and comprehensively explore this effect in relation to other factors that may have influenced detection rates within an information-theoretic context. Substrate also emerged as a factor affecting detection on the ACUA site, an occurrence that has been demonstrated in other studies but is not always accounted for (Morrison 2002; Johnson et al. 2004; Arnett et al. 2008). Carcasses located on bare ground or pavement were significantly more likely to be detected than were those on gravel, and our detection models provided parameter estimates for this substrate effect. Our study also confirmed the assumption that detection rates can vary among observers (Morrison et al. 2001; Morrison 2002), and that such differences can be significant, but does not imply that observer effects are prevalent in all studies. Note that the observer differences reported here are specific to the observers in this study only, and could change over time even if the observers remain the same.
Known-fate survival models showed that carcass scavenging rates declined over time, with most carcasses removed by scavengers within 1–3 d of initial placement. Scavenging rates appeared to stabilize to approximately 2%/d after 5 d, so that after 14 d 31% of carcasses would be predicted to remain. The general pattern we observed supports the assumption of exponential decrease in scavenging rates over time since carcass placement (due to decomposition and desiccation), as has been proposed by Erickson (2006), and demonstrated by Smallwood et al. (2010), although the shape of this curve likely varies by site and season (see also Smallwood et al. 2010; Korner-Nievergelt et al. 2011; Bispo et al. 2013). Although our limited sample size precluded us from making inferences about the relationship between season and temporal patterns in scavenging rates, seasonal effects have been documented elsewhere with varied results, indicating that seasonal patterns likely differ among sites and regions (Osborn et al. 2000; Smallwood 2007; Smallwood et al. 2010; Bispo et al. 2013). Peron et al. (2013) propose a method to account for the potential effects of carcass age on scavenging rates when estimating wind-power-caused mortality and its associated level of uncertainty.
The detection and scavenging parameter estimates identified through the modeling processes described in this study generated empirically derived, unbiased estimates of searcher efficiency and removal probability. These estimators can subsequently be used to determine the likelihood that each wind-caused mortality is 1) available and 2) detectable, which are the two parameters that comprise the core bias-adjustment factors in most mortality models. Methods commonly used for estimating efficiency and scavenging rates have generally relied on assumptions that were not fully supported by our findings. For instance, differences in carcass detectability were acknowledged in an earlier report summarizing the ACUA data presented here, but were addressed by using mean raw values (i.e., mean efficiency rates for small birds, large birds, and bats; Tidhar et al. 2011) in efficiency estimation models. This approach did not take into account the observer or substrate differences we found to be inherent in the data, nor the uncertainty associated with the size-bias estimates, which could greatly affect subsequent modeling results (McGowan et al. 2011). Furthermore, it has been argued that many estimators yield scavenging parameters that are highly sensitive to study duration and can thus be severely biased based on arbitrarily defined study periods (e.g., mean time to removal, proportion of carcasses remaining, Smallwood 2007; Korner-Nievergelt et al. 2011), whereas our estimates were based on an empirical persistence curve that was not dependent on trial duration. Overall, the temporal changes in scavenging rates observed in this study justify prior concerns that assumptions about scavenging rate distributions might be violated and could severely affect subsequent correction factors (Bispo et al. 2013).
Increasing the confidence in, and precision of, detection and scavenging estimates at individual wind-assessment sites would greatly improve confidence in mortality estimates (e.g., Korner-Nievergelt et al. 2011; Peron et al. 2013; Stevens and Dennis 2013). Identifying other potential sources of detection bias that were not accounted for in this study can also be achieved through the methods outlined herein (e.g., Bispo et al. 2013). For instance, the effects of using frozen vs. fresh trial carcasses may affect scavenging-rate estimates, an assumption that should be more rigorously field-tested. Carcass age also likely has an effect on detectability (i.e., is there a threshold age at which carcasses are so dessicated they are essentially undetectable?), and estimates of the rate of decrease could be verified using known-age carcasses. We suggest that future wind-mortality assessments incorporate methods like those described in this study to identify site-specific sources of variation in detection and scavenging rates, and embed results in likelihood methods that use search trials and carcass survey data to estimate total mortality (e.g., Peron et al. 2013).
As wind-power continues to rapidly expand on a continental and global scale, understanding its effect on breeding, overwintering, and migrating birds and bats will become increasingly important. Generating reliable estimates of detection, scavenging, and resultant mortality estimates will be extremely valuable for guiding management geared toward mitigating or avoiding negative impacts. The National Research Council (NRC 2007) has expressed concern about how wind power affects population drivers such as genetic structure and demographics for species at risk, although questions about the linkage between avian population health and collision mortality remain (Arnold and Zink 2011). By incorporating standardized methods for generating unbiased mortality estimates into current wind-power assessment guidelines, these and other uncertainties can be more reliably addressed.
Please note: The Journal of Fish and Wildlife Management is not responsible for the content or functionality of any supplemental material. Queries should be directed to the corresponding author for the article.
Table S1. Data file characteristics for data obtained during carcass detection trials conducted at the Jersey Atlantic Wind, LLC (JAW)/Atlantic City Utilities Authority (ACUA) wind power facility near Atlantic City, New Jersey, September 2007–June 2010. Variables are Carcass ID (ID), Species (SP), Taxon (1 = Bird, 0 = Bat), Size (1 <35g, 0 >35g), Mass (g), Substrate (Bare Ground, GraVel, GRass), Searcher, Temp (F), and Detected (1 = Yes, 0 = No). Detailed data descriptions are provided in accompanying text file (readme.txt).
Found at DOI: http://dx.doi.org/10.3996/032014-JFWM-019.S1 (42 KB RTF).
Table S2. Data file characteristics for data obtained during 7-d scavenging trials conducted at the Jersey Atlantic Wind, LLC (JAW)/Atlantic City Utilities Authority (ACUA) wind power facility near Atlantic City, New Jersey, September 2007–June 2010. Variables are Carcass ID (ID), carcass Encounter History (Cooch and White 2011), Spring (1 = Yes, 0 = No), Summer (1 = Yes, 0 = No), Autumn (1 = Yes, 0 = No), Taxon (bird = 1, bat = 0), AvgMass (average species mass [g]), Distance from Marsh (m) and AvgTemp (mean temperature [F] at Atlantic City International Airport during trial period). Detailed data descriptions are provided in accompanying text file (readme.txt).
Found at DOI: http://dx.doi.org/10.3996/032014-JFWM-019.S1 (42 KB RTF).
Supplemental Data S1. Data file.
Found at DOI: http://dx.doi.org/10.3996/032014-JFWM-019.S2 (24 KB XLSX).
Reference S1. [ABC] American Bird Conservancy. 2011. Rulemaking petition to the U.S. Fish & Wildlife Service for regulating the impacts of wind energy projects on migratory birds. 14 December 2011.
Found at DOI: http://dx.doi.doi.org/10.3996/032014-JFWM-019.S3; also available at http://www.abcbirds.org/abcprograms/policy/collisions/pdf/wind_rulemaking_petition.pdf (1814 KB PDF)
Reference S2. [ACUA] Atlantic County Utilities Authority. 2014. Jersey-Atlantic wind farm fact sheet.
Reference S3. Erickson WP 2006. Objectives, uncertainties and biases in mortality studies at wind facilities. Paper presented to the National Wind Coordinating Collaborative (NWCC) Research Meeting VI, November 2006. San Antonio, Texas.
Found at DOI: http://dx.doi.org/10.3996/032014-JFWM-019.S5; also available at http://nationalwind.org/wp-content/uploads/assets/research_meetings/Research_Meeting_VI_Proceedings.pdf (1814 KB PDF)
Reference S4. Jain A, Kerlinger P, Curry R, Slobodnik L. 2007. Annual report for the Maple Ridge Wind Power Project postconstruction bird and bat fatality study - 2006. Unpublished report submitted to PPM Energy and Horizon Energy and Technical Advisory Committee (TAC) for the Maple Ridge Project, February 2007. Cape May Point, New Jersey: Curry & Kerlinger, L.L.C.
Found at DOI: http://dx.doi.doi.org/10.3996/032014-JFWM-019.S6 (1814 KB PDF)
Reference S5. Johnson GD, Erickson WP, Strickland MD, Shepherd MF, Shepherd DA. 2000. Avian Monitoring Studies at the Buffalo Ridge Wind Resource Area, Minnesota: results of a 4-year study. Technical report prepared by WEST, Inc. for Northern States Power Co., Minneapolis, Minnesota.
Found at DOI: http://dx.doi.doi.org/10.3996/032014-JFWM-019.S7 (1814 KB PDF)
Reference S6. Kerns J, Erickson WP, Arnett EB. 2005. Bat and bird fatality at wind energy facilities in Pennsylvania and West Virginia. Pages 24–95 in Arnett EB, editor. Relationships between bats and wind turbines in Pennsylvania and West Virginia: an assessment of bat fatality search protocols, patterns of fatality, and behavioral interactions with wind turbines. A final report submitted to the Bats and Wind Energy Cooperative. Austin, Texas: Bat Conservation International.
Found at DOI: http://dx.doi.doi.org/10.3996/032014-JFWM-019.S8 (1814 KB PDF)
Reference S7. Mizrahi DS. 2009. Avalon Seawatch: long-term monitoring water bird migration along the southern Atlantic Coast of New Jersey. Annual report submitted to Sea Duck Joint Venture
Reference S8. Morrison ML. 2002. Searcher bias and scavenging rates in bird-wind energy studies. Golden, Colorado: National Renewable Laboratory Report NREL/SR-500-30876.
Reference S9. Tidhar D, Nomani S, Erickson W. 2011. Post-construction bat and bird multi-year fatality estimate comparison, ACUA Wind Resource Area. WEST, Inc. report submitted to Bluarc Management Group/Atlantic County Utilities Authority, Atlantic City, New Jersey.
Found at DOI: http://dx.doi.doi.org/10.3996/032014-JFWM-019.S11 (1814 KB PDF)
Reference S10. [USFWS] U.S. Fish and Wildlife Service. 2012. U.S. Fish and Wildlife land based wind-energy guidelines. OMB Control No, 1018-0148, March 2012.
Found at DOI: http://dx.doi.doi.org/10.3996/032014-JFWM-019.S12 (1814 KB PDF)
Reference S11. Wiser R, Bolinger M. 2012. 2011 Wind technologies market report. Report submitted to Office of Energy Efficiency and Renewable Energy of the U.S. Department of Energy, DE-AC02-05CH11231.
Funding for this project was provided by Community Energy and Jersey Atlantic Wind, LLC. We received valuable input on earlier versions of this manuscript from Jim Nichols, Guillaume Peron, and two anonymous reviewers.
Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government.
Peters KA, Mizrahi DS, and Allen MC. 2014. Empirical evidence for factors affecting searcher efficiency and scavenging rates at a coastal, terrestrial wind-power facility. Journal of Fish and Wildlife Management 5(2):330–339; e1944-687X. doi: 10.3996/032014-JFWM-019
The findings and conclusions in this article are those of the author(s) and do not necessarily represent the views of the U.S. Fish and Wildlife Service.