Introduction

One of the bigger and often the biggest environmental problem caused by wind farms is the direct mortality of flying fauna (mainly birds and bats) inflicted by wind turbines. For example, it is estimated that the population of Lasiurus cinereus (the bat species most frequently killed by wind turbines in North America) may decline by as much as 90% in the next 50 years due to mortality from wind farms [11]. For this reason, it is critically important to adequately conduct a previous study of the bird and bat populations of the site in order to evaluate the environmental impact of the project before its construction as well as a mortality monitoring throughout the entire period of operation of the wind farm.

The present guidelines summarises the critical aspects needed to carry out a proper environmental impact assessment of wind farms. It is the result of the author's own experience and the analysis of the numerous articles as well as recommendations and guidelines that have proliferated all over the planet.

Study of wildlife populations

In order to assess the mortality of flying animals, the unit of study must be each wind turbine and more specifically each possible wind turbine location (in order to have data that allow comparisons between possible alternatives of wind turbine location). This is important because mortality can vary significantly between wind turbines of the same wind farm, so studying only part of the wind turbines can lead to erroneous conclusions.

The study must cover at least a whole year, due to the significant changes in the use of space for each species throughout the annual cycle.

The study of birds must be based on observation stations. An observation station must be placed at the location of each possible wind turbine. In order to get a good view of the environment, the observer can move up to a maximum distance (from the central point of the wind turbine) equal to half the maximum height of the wind turbine (height above ground plus radius of the rotor). Observation time at each station must be 30 minutes per week, that can be distributed over some observation periods. The work must be carried out exclusively by people with sufficient knowledge and experience for visual and auditory identification of all species likely to be present in the area. The field work must collect for each observation all the data necessary for the subsequent application of a collision risk model (mainly time in the risk zone, which is defined as time spent flying at a height within the turning range of the wind turbine blades and at a distance equal to or less than 200 meters from the positions of the wind turbines).

In the case of bats, ideally an ultrasonic recording equipment with a microphone at rotor height and another microphone near the ground (less than 20 meters high) (and additionally thermal or infrared image recording equipment at rotor height if non-echolocalizator bat species are present) should be installed at the location of each possible wind turbine, but this proposal is usually not feasible during the pre-construction phase. Otherwise, equipments of the indicated characteristics must be placed where there are facilities that allow it (usually the mast(s) or tower(s) of the meteorological stations), complemented by night visits to the location of each wind turbine if possible or, failing that, to the closest locations that the road network allows, carrying an ultrasound detector and counting 30 minutes per week per wind turbine. Only detectors in direct ultrasonic recording mode and with a microphone with highly sensitive in the entire frequency range used by the potentially present species must be used exclusively. The status of the microphones must be monitored, deteriorated microphones should not be used. The identification of the recordings must be carried out by people with sufficient knowledge and experience, this task should not be delegated to automatic identification systems due to their low reliability [4, 9].

Estimate of expected mortality

The data of observations in risk zone must be used to estimate the expected direct mortality using collision risk models (because there is no simple relationship between abundance and mortality), mainly the Scottish Natural Heritage model developed by William Band and collaborators. When interpreting the results, it should be taken into account that, as with any model of this nature, the final results are very sensitive to the quality of the input data, and in this specific case the two most significant variables [3] are: the numerical values ​​of time in the risk zone obtained from field sampling, and the "evasion rates" used to adjust the raw results of the model to reality.

The results of estimated mortality must be used to evaluate the specific impact on the populations of each species (with special attention to threatened or scarce ones) through the use of demographic models (population viability analysis), given that what is critical is not so much the overall mortality numbers but rather the impact on species population, since a small number of deaths can seriously affect scarce, threatened or sensitive species.

Additionally, as an alternative way of estimating impact, spatial vulnerability indices [5, 8] must also be applied.

Monitoring of mortality

When studying the direct mortality of a wind farm on flying vertebrate fauna (birds and bats), it is critical to keep in mind that the mortality detected by searching for carcasses on the field is only a fraction of the real mortality. On the one hand, a part of the accidents do not cause death in situ, but rather the produced injuries allow the animal to fly away and die at a certain distance from the wind farm in the following minutes or hours (ex situ mortality). On the other hand, from the moment that the body falls to the ground, scavengers, decomposers and meteorological agents begin to act causing its disappearance. And finally, the efficiency of carcasses detection by technical personnel by the technical personnel is not perfect, and frequently within the search area there are different vegetal covers with different carcass detectability. Based on this, the following relationship can be established:

R ~ E, C, P, D

Where:

• R = Real mortality.
• E = Ex situ mortality.
• C = Found carcasses.
• P = Temporal rate of disappearance of carcasses.
• D = Efficiency of carcasses detection by technical personnel.

The D (efficiency of carcasses detection by technical personnel) and P (temporal rate of disappearance of carcasses) variables vary depending on factors specific to the carcass (such as its size), the environment (characteristics of the vegetation cover), and the dynamic agents (variations throughout the year in the activity of scavengers, decomposers, and meteorological agents), and in the first case, also on the technical personnel.

A general model to calculate the estimates of real mortality is:

C_k,l,m,n = R_k,l,m,n · ( 1 – E_k ) · P(t)_k,l,m · D_k,l,m,n · F_l

Where:

• P(t) = Average proportion of carcasses which persists for a time equal to half of the time interval between visits ( t = Δt_i / 2 ).
• k = Referring to the k species, or otherwise the k species group defined according to their similarity in size.
• l = Referring to each different vegetal cover existing on the search area.
• m = Referring to each differentiable period of the year depending on variations in the temporal rate of carcass disappearance and the efficiency of carcasses detection by technical personnel.
• n = Referring to the efficiency of carcasses detection by technical personnel.
• F = Extrapolation factor defined as the ratio between the sampled area and the area inside the circle with centre in the wind turbine and radius equal to the maximum height of the wind turbine (height above ground plus radius of the rotor).

Thus, for the wind turbine x_i in the interval of time between visits   Δt_i = t_i – t_i–1  the estimated total real mortality is equal to the summation of the estimated partial real mortalities for the different combinations of k, l, m, and n

R_(xi,Δti) = Σ_(xi,Δti) R_k,l,m,n

The efficiency of carcass detection by technical personnel (D) and the temporal rate of carcass disappearance (P) must be estimated experimentally with appropriate sample size, randomness, and frequency (at least once in each season of the year) for the different combinations of factors: species or group of species (k), vegetation cover (l), period of the year (m), and technical personnel conducting the search (n). To do this, a person must distribute and geolocate (adequately encoding each one and using satellite positioning: GPS, GLONASS,...) carcasses of wild birds and bats of different sizes (from deaths in wind farms, air power lines and roads) and animals bred in captivity such as mice, quails, hens/roosters and other cage and poultry birds of brown colours, in the search areas around the wind turbines, over as many days as possible (in order to avoid accumulations that alter the behaviour of the scavengers), and using gloves (both for health reasons and to avoid impregnating abnormal odors). In the case of efficiency of carcass detection by technical personnel (D), in order to avoid the action of scavengers and to avoid the possible "feel under examination" effect, the person who distributes the carcasses must leave the area shortly before the arrival of the tested technical personnel (who should not know that it is an experiment day to ensure that it is considered a routine carcass search visit) and also the subsequent the subsequent collection and check of whether undetected carcasses remain where they were deposited or have been scavenged should be carried out as soon as possible. In the case of the temporal rate of carcass disappearance (P), only carcasses in good condition, not rotten, and well thawed will be used, and they must be placed halfway between visits (3-4 days after the previous visit and 3-4 days before the next visit if the frequency of visits is weekly), so that the routine carcass search visit serves to determine the disappearance rate for a time equal to half of the time interval between visits ( t = Δt_i / 2 ). Because of the significant differences found occasionally between close locations, the values ​​obtained in nearby wind farms must not be used.

If any vegetation cover cannot be sampled or its carcass detection efficiency (D) is zero, the real mortality is estimated for the remaining proportion of the search area and the result is extrapolated to the total. Since extrapolations should be avoided in order to achieve the most accurate estimates, it is highly recommended that an area around each wind turbine with a radius equal to the maximum height of the wind turbine (height above ground plus radius of the rotor) be kept free of vegetation or only with low vegetation (<10 cm high), because it greatly increases the efficiency of carcass detection allowing much better real mortality estimates, while revegetation has little positive environmental effect (and can be negative in agricultural environments by attracting wildlife).

Due to insufficient data, it is tentatively suggested to use value of ex situ mortality (E) equal to 10% (0.1).

Since mortality can vary significantly between wind turbines of the same wind farm and considering the usual values of temporal rate of disappearance of carcasses, the carcass searching must be carried out on all wind turbines with weekly as minimum frequency.

Ideally the search area around each wind turbine should be a square with center in the position of each wind turbine and apothem equal to the maximum height of the wind turbine (height above ground plus rotor radius). However, due to the tendency to install increasingly large wind turbines, the search area per wind turbine can be quite large which can cause a counterproductive reduction in the efficiency of carcasses detection by technical personnel. As an alternative, it is possible to conduct the carcass search within a square of minor apothema and extrapolate the results, for example an apothem equal to 75% of the rotor radius [1]. The square rather than spiral design facilitates the performance of a linear zig-zag transect in parallel bands of maximum 5 meters wide (2.5 meters on each side, twice as wide in the case of dogs) at a maximum speed of 1 m/s. In case of using dogs, they will carry a satellite tracking device that accurately records the track done.

There are two methods for carcass searching: by people or by dogs. Dogs tend to offer higher detection efficiency values but with greater uncertainty and less accuracy, and greater time efficiency than people. The problem with dogs is that it involves considering numerous additional factors that are difficult to estimate and control its variability and correlation between them: the variability of different combinations of dog and person teams, the variation of the dog's mood between days and throughout the day, the variability of detectability between vegetation cover from the canine perspective that is not identified by people, the different detectability between bird and bat species, the variability of olfactory detectability according to the state of decomposition of the carcasses, and the effect of local meteorological factors such as temperature, humidity, wind speed and direction, in addition to a greater difficulty in estimating the efficiency of carcasses detection due to the problem of permeating abnormal odors in the carcasses used in the experiments (causing an overestimation of detection efficiency that leads to an underestimation of real mortality). The advantage of dogs is that a greater number of found carcasses provides a greater numerical basis for analysis and a lower probability of false zeros (no detection of existing carcasses). Whenever possible, it is recommended that both methods be tested for each wind farm (alternating wind turbines and alternating visits), comparing the results and concluding on the basis of the results obtained. It is highly recommended to keep only herbaceous vegetation or no vegetation around the wind turbines, because it greatly increases the efficiency of carcasses detection allowing much better estimates of real mortality, while revegetation has little positive environmental effect (and can be negative in agricultural environments by attracting wildlife).

The data obtained on detected mortality and estimated real mortality should be compared with the results of the estimates of expected mortality in order to improve collision risk models and spatial vulnerability indices and recalculate population viability analyzes.

Wildlife protection measures

Preventive measures are always the best option. And the best preventive measure is not to install a wind farm where a serious affection to scarce, threatened or sensitive species is expected (as guide numbers: <5 km of nests for medium birds, <15 km of nests for large or scavenger birds, <5 km of shelters for bats).

In the case of bats, the only measure that can be recommended to reduce mortality is to stop the wind turbines at night with wind speed ≤6 m/s and temperature ≥10 ºC (somewhat lower value for some species [7]) which is a threshold that has proven to be effective, scientifically confirmed and economically acceptable [2, 6].

In the case of birds, a cheap and effective measure is that wind turbines are dark in colour to improve their visibility instead of white.

Automatic detection and collision prevention systems using radar or video [10] allow reducing bird mortality, but with limited effectiveness, at an affordable price. Because the price of a video-based system is 3-4% of a wind turbine, while that of a radar-based system is 70-75%, the former are the most widely used except when the mortality problem is concentrated at night (migration or bats) or in frequent foggy conditions. Only wind turbine stop modes are recommended and conditioned to be configured to allow the effective stopping of the wind turbines in time to avoid collisions depending on the distance and speed of the birds, the use of acoustic signals is usually discouraged due to their low efficiency in avoiding collisions and the problems of disturbance it leads.

Bibliographic references:

[1] Cindy Hull, Sheldon Muir. Search areas for monitoring bird and bat carcasses at wind farms using a Monte-Carlo mode. Australasian Journal of Environmental Management, 17(2) (2010).

[2] Colleen M. Martin, Edward B. Arnett, Richard D. Stevens, Mark C. Wallace. Reducing bat fatalities at wind facilities while improving the economic efficiency of operational mitigation. Journal of Mammalogy, 98 (2): 378–385 (2017).

[3] Dan Chamberlain, Steve Freeman, Mark Rehfisch, Tony Fox, Mark Desholm. Appraisal of Scottish Natural Heritage's wind farm collision risk model and its application. British Trust for Ornithology, Research Report 401 (2005).

[4] Jens Rydell, Stefan Nyman, Johan Eklöf, Gareth Jones, Danilo Russo. Testing the performances of automated identification of bat echolocation calls: A request for prudence. Ecological Indicators, 78: 416-420 (2017).

[5] José C. Noguera, Irene Pérez, Eduardo Mínguez. Impact of terrestrial wind farms on diurnal raptors: developing a spatial vulnerability index and potential vulnerability maps / Impacto de campos eólicos terrestres sobre rapaces diurnas: desarrollo de un índice de vulnerabilidad espacial y mapas de vulnerabilidad potencial. Ardeola, 57(1): 41-53 (2010).

[6] Oliver Behr, Robert Brinkmann, Klaus Hochradel, Jürgen Mages, Fränzi Korner-Nievergelt, Ivo Niermann, Michael Reich, Ralph Simon, Natalie Weber, Martina Nagy. Mitigating bat mortality with turbine-specific curtailment algorithms: a model based approach. Wind Energy and Wildlife Interactions: 135-160 (2017).

[7] Raphaël Arlettaz, Catherine Ruchet, John Aeschimann, Edmond Brun, Michel Genoud, Peter Vogel. Physiological traits affecting the distribution and wintering strategy of the bat Tadarida teniotis. Ecology, 81(4): 1004-1014 (2000).

[8] Roberto Toffoli, Paola Culasso, Paolo Oberto. Wind farms and preventive evaluation of impacts on bats: a case study. II Convegno Internazionale Fauna Problematica: Conservazione e Getione, Genazzano (Italia) (2011).

[9] Robin Brabant, Yves Laurent, Umit Dolap, Steven Degraer, Bob Jonge Poerink. Comparing the results of four widely used automated bat identification software programs to identify nine bat species in coastal Western Europe. Belgian Journal of Zoology, 148 (2): 119-128 (2018).

[10] Sjoerd Dirksen. Review of methods and techniques for field validation of collision rates and avoidance amongst birds and bats at offshore wind turbines. Sjoerd Dirksen Ecology (2012).
Technische Systeme zur Vermeidung von potenziellen Auswirkungen auf Vögel und Fledermäuse durch die Windenergienutzung. Kompetenzzentrum Naturschutz und Energiewende (2018).
Eva Schuster, Elke Bruns. Technische Ansätze zur bedarfsgerechten Betriebsregulierung. Naturschutz und Landschaftsplanung, 50 (7): 226-232 (2018).
島田泰夫.風力発電事業で懸念される影響の回避･低減に関する技術--鳥類を中心に--. 日本気象協会 (2019).

[11] Winifred Frick, Erin Baerwald, Jacob Pollock, Robert Barclay, Jennifer Szymanski, Theodore Weller, Amy Russell, Susan Loeb, Rodrigo Medellín, Rodrigo Medellin. Fatalities at wind turbines may threaten population viability of a migratory bat. Biological Conservation, 209: 172-177 (2017).