Scottish Marine and Freshwater Science Volume 5 Number 16:The Avoidance Rates of Collision Between Birds and Offshore Turbines
This study reviewed data that have been collected from offshore windfarms and considers how they can be used to derive appropriate avoidance rates for use in the offshore environment.
Appendix 6 Evidence review within-windfarm avoidance
A6.1 Avonmouth Docks
The Landmark Practice. 2013. Birds and Wind Turbines At Avonmouth Docks. Year 5 Monitoring Report for Ecotricity. The Landmark Practice, Bristol
Methods
Monitoring was undertaken at the Avonmouth Docks windfarm between October and March in the winters of 2007/08, 2008/09, 2009/10, 2011/12. Three vantage point surveys, each lasting three hours, were carried out in each month to record bird activity at the site. Flight altitude was estimated in five bands 0-20 m, 20-40 m, 40-80 m, 80-160 m and >160 m.
During the visits for each vantage point survey, a search with a radius of 60 m around each turbine was carried out for corpses. Additional surveys were carried out following periods of severe weather. In total 343 checks were carried out around the base of each turbine in the post-construction period.
Seasons / time of day
Vantage point surveys were carried out between October and March, and timed so that periods of rising, falling and high tide were covered each month.
Species
Black-headed gull (4.4 birds/hr 2007/08, 7.1 birds/hr 2008/09, 2.9 birds/hr 2009/10, 12.8 birds/hr 2011/12), herring gull (6.8 birds/hr 2007/08, 13 birds/hr 2008/09, 18.8 birds/hr 2009/10, 38.2 birds/hr 2011/12)
Conditions data collected under
Not stated.
Location / habitat
Avonmouth Docks, coastal.
Turbine / array specification
A line of 3, 2 MW Enercon E82 turbines, with a hub height of 79m and a rotor diameter of 83 m.
Results
A single black-headed gull was identified as a probable collision victim in the winter of 2007/08. An average of 4.4 black-headed gulls were recorded passing through the site over the study period, suggesting a total flux rate of 10,530 birds, of which 57 were predicted to collide based on option 1 of the Band model, 2 were predicted to collide based on option 2 of the Band model and 1 was predicted to collide based on option 3 of the Band model. This reflects avoidance rates of 0.9826 using option 1 of the Band model, 0.5152 using option 2 of the Band model and -0.0005 using option 3 of the Band model.
Assessment of methodology
The corpse search methodology is likely to provide an accurate estimate of collision numbers as previous studies have shown that the majority of corpses are recovered within 40 m of a turbine base (Orloff & Flannery 1992, Munster et al. 1996, Howell 1997). Furthermore, corpses were examined to confirm collision as cause of death. No corrections were carried out to account for searcher efficiency or predator activity. However, given the habitat surrounding the turbines and the frequency of searches through the study period, it is unlikely corpses would have been missed. Bird activity surveys were carried out throughout the study period and are therefore likely to give a realistic impression of bird activity in the area.
As the bird activity surveys were carried out concurrently with the corpse searches and covered the same area, these data were combined with data from other sites to estimate representative avoidance rates.
A6.2 Altamont Pass
Thelander, C.G., Smallwood, K.S. & Rugge, L. 2003. Bird risk behaviours and fatalities at the Altamont Pass Wind Resource Area. National Renewable Energy Laboratory, Colorado.
Methods
Visual observations and fatality searches.
Circular areas with a 50 m radius around the base of 685 wind turbines were searched for corpses every five to six weeks between 1998 and 2000. These searches were combined with 1,958 30 minute point counts carried out in 20 study plots on 303 different days between 1998 and 2000.
Seasons / time of day
Counts carried out throughout the year and between 0700 h and dusk.
Species
Gulls (0.48 birds/hour).
Conditions data collected under
All conditions unless wind or rain resulted in visibility dropping to <60 m.
Location / habitat
Altamont Pass, California, U.S.A. (terrestrial).
Turbine / array specification
685 turbines arranged in 109 rows across an area of 50 km 2. Turbine hub heights ranged from 14 m-30 m, with rotor diameters of 17-23 m. Distance between turbines within each row is not described.
Results
At this site, a total of five gulls, of unknown species, were recovered following collision with turbines. Across the study plots as a whole, the average rate at which gulls passed through the windfarm was 0.48 birds per hour, reflecting a total of 7,428 gull movements within the area over the two year study period. Site specific flight height data were not available, so it was not possible to calculate an avoidance rate based on option 1 of the Band model. Assuming no avoidance behaviour, and a bird with the characteristics of a herring gull, the total number of collisions expected would have been 296 per annum under option 2 of the Band model and 295 under option 3 of the Band model. The collision rate of five birds over the study therefore indicates a within-windfarm avoidance rate of 0.9831 using option 2 and 0.9831 using option 3. The similarity between these values reflects the relatively small size of the turbines installed at the site, in particular the rotor-swept area, diameters of 17-23 m are significantly smaller than many of the turbines installed at offshore sites.
Assessment of methodology
The corpse search methodology is likely to provide an accurate estimate of collision numbers as previous studies have shown that the majority of corpses are recovered within 40 m of a turbine base (Orloff & Flannery 1992, Munster et al. 1996, Howell 1997). Furthermore, corpses were examined to confirm collision as cause of death. Correction factors were applied to account for carcass removal by scavengers, but not to correct for searcher efficiency. However, the limited size of the search area and terrain made it unlikely that any corpses would have been undetected.
To minimise the effects of observer bias in point counts, paired observations were carried out during the early part of the study period so that different observers calibrated their perceptions of altitude, distance and behaviour with one another. However, no correction was applied for the detection distance of different species. This is a concern given that study plots were up to 4 km 2, meaning that the total number of birds present within the study areas may have been an underestimate and that, therefore, the final, derived avoidance rate would also have been an underestimate.
However, as it has been necessary to extrapolate bird activity data across the site, this has not been combined with data from other sites to identify representative avoidance rates.
A6.3 Blyth Harbour
Visual observations and fatality searches.
Lawrence, E.S., Painter, S. & Little, B. 2007. Responses of birds to the wind farm at Blyth Harbour, Northumberland, UK. In de Lucas, M., Janss, G.F.E., Ferrer, M. 2007. Birds and Wind farms Risk Assessment and Mitigation, Quercus, Madrid.
Newton, I. & Little, B. 2009. Assessment of wind farm and other bird casualties from carcasses found on a Northumbrian beach over an 11-year period. Bird Study, 56, 158-167.
Methods
Once a week over an 11 year period, a 4.7 km stretch of beach near Blyth in Northumberland was searched for corpses. Depending on the condition of the birds, an attempt was made to assign a cause of death to each carcass, and those with symptoms thought to be typical of collision with a wind turbine - head or one or both wings missing, broken bones blood in body cavity and a ruptured liver - were identified. The total number of carcasses found was then corrected to account for those lost to scavengers, those not washed up on the beach and those not found during searches.
Between October 1996 and August 1998, 31 three hour-long periods of observation were made of flight activity perpendicular to the turbine row and in the vicinity of five of the nine turbines. Observations were made from a point on the shore opposite the turbines, at a distance of approximately 80 m. In total 93 hours of observational data were collected.
Seasons / time of day
Fatality data were collected throughout the year. Bird activity data were also collected throughout the year, between the hours of 0800 and 1500 h, with observation periods split equally between the morning and afternoon.
Species
Around 80% of the flight activity within the windfarm involved herring gull and great black-backed gull, and other gull species made up a significant proportion of the remaining species. However, as species-specific data were not available regarding the corpses collected and it was stated that the majority of those collected belonged to gulls, to calculate an avoidance rate, it was necessary to consider gulls collectively.
Conditions data collected under
No details given.
Location / habitat
Blyth Harbour breakwater, Northumberland, UK (coastal).
Turbine / array specification
Nine turbines arranged in a row along a harbour breakwater. The turbines are spaced at 200 m intervals and have a hub height of 25 m with a 25 m rotor diameter.
Results
Results were presented as average collision rates and passage rates over the study period as a whole. Based on the data presented an average of 417,954 birds, most of which were large gulls, would have been expected to pass through the windfarm over the study period. Of these, approximately 3,047, assuming birds with the characteristics of a herring gull, would have been expected to collide with turbines in the absence of avoidance behaviour using option 1 of the Band Model and 3,083 using option 2 and 3,007 using option 3. Having corrected for the imperfect detection of corpses, between 148.5 and 193.5 collisions with wind turbines were expected in an average year. This suggests a within-windfarm avoidance rate of 0.3966-0.5369 using option 1, 0.4037-0.5423 using option 2 and 0.3886-0.5308 using option 3.
Assessment of methodology
The fatality searches were intensive throughout the study period and followed a robust methodology to account for corpses that went undetected. In particular, the potential for corpses to wash up within the study area was tested experimentally.
The observational data were limited to a two year period in the middle of the study. The data may have underestimated gull movements within the surrounding area for two key reasons. Firstly, no corrections were applied to account for imperfect detection of birds. Secondly, by limiting observations to the period between 0800 and 1500 h, key movements of gulls to and from roost sites may have been missed during the summer and autumn. Underestimating bird activity within the area would lead to an underestimate of the number of collisions expected in the absence of avoidance behaviour, and consequently, the final derived avoidance rates would also be underestimated.
Activity data were only collected between 206 and 2008 and only between turbines 5 and 9. As the mean annual collision rates relate to the whole of the study period, and to all 9 turbines, it is necessary to extrapolate activity data both temporally and spatially to derive a flux rate. Therefore, these data have not been included when deriving representative avoidance rates.
A4.4 Blyth Offshore Windfarm
Rothery, P., Newton, I. & Little, B. 2009. Observations of seabirds at offshore wind turbines near Blyth in northeast England. Bird Study, 56, 1-14
Methods
Visual observations
Following the installation of the offshore turbines, observations of birds in the vicinity of the turbines were made on 177 occasions between 12 January and 30 August 2003, totalling almost 352 hours of observation. Observations were made from the shore and distances and heights of flying birds were calibrated against objects of known size and fixed locations.
Seasons / time of day
Observations were made between January and August. Data collection was focussed on the period between 1130 and 1600, consequently, during the summer movements to and from breeding colonies may have been missed.
Species
Northern gannet (1.23 birds/hr), great cormorant (1 bird/hr), common scoter (0.96 birds/hr), common eider (2.77 birds/hr), black-headed gull (2.78 birds/hr), herring gull (4 birds/hr), great black-backed gull (1.6 birds/hr), black-legged kittiwake (3.83 birds/hr), Sandwich tern (6.07 birds/hr).
Conditions data collected under
No details given.
Location / habitat
Blyth, Northumberland, UK (offshore).
Turbine / array specification
Two turbines separated by 200 metres. Each turbine had a hub height of 59.4 m above mean sea-level and a rotor diameter of 66 m.
Results
Throughout the study period, no collisions were recorded involving any of the species observed in the vicinity of the windfarm, reflecting a within-windfarm avoidance rate of 1.0000 for each species considered (Northern gannet, great cormorant, common scoter, common eider black-headed gull herring gull, great black-backed gull, black-legged kittiwake and Sandwich tern).
Assessment of methodology
No corrections were applied to account for the imperfect detection of birds during the survey. Consequently, the true level of bird activity within the study area was likely to have been underestimated. Additionally, it was not possible to search for carcasses, meaning that inferences about avoidance behaviour can only be drawn from the failure of observers to detect a collision from a total of 352 hours of monitoring. Given the low probability of a collision occurring, and the levels of flight activity recorded, this outcome is unsurprising. It is also important to note that the size of the OWF was very small (two turbines) and therefore caution must be applied when considering how applicable these avoidance rates are for much bigger arrays.
As insufficient observational data have been collected to record a collision, these data have not been included in those used to derive representative avoidance rates.
A6.5 Boudwijnkanaal
Everaert, J. 2008. Effecten van windturbines op de fauna in Vlaanderen: Onderzoeksresultaten, discussie en aanbevelingen. INBO, Brussels
Everaert, J. & Kuikjen, E. 2007. Wind turbines and birds in Flanders (Belgium): Preliminary summary of the mortality research results. INBO, Brussels
Everaert, J., Devos, K. & Kuijken, E. 2002. Windturbines en vogels in Vlaanderen: Voorlopige onderzoeksresultaten en buitenlandse beviningen. INBO, Brussels
Methods
Visual observations and fatality searches.
Systematic fatality searches were carried out once every 14 days between 2001 and 2006. Searches were carried out within a circular area, with a radius of 100 m, centred on each turbine. Corrections were applied to the data to account for imperfect detection and searcher efficiency.
Observational data describing the number of birds passing the turbine hub were collected between September and December 2005 between turbines 8 and 14. The resultant data were used to extrapolate the total number of birds likely to have passed the turbines over this period. Observational data are presented as a mean daily total collected during the period from two hours before dawn to four hours after dusk in October, reflecting a total of 17 hours of observations.
Seasons / time of day
Fatality data were collected throughout the year, behavioural data were collected between September and December.
Species
Gulls (1,075 birds/day).
Conditions data collected under
No details given.
Location / habitat
Boudwijnkanaal, Brugge, Belgium (terrestrial).
Turbine / array specification
A row of 14 turbines, each with a hub height of 55 m and a rotor diameter of 48 m. Distance between turbines within each row is not described.
Results
Collisions involving gulls were recorded in each year of the study, with a minimum of 21.2 collisions occurring in 2001 when only five of the 14 turbines were operational and a maximum of 264.6 collisions occurring in 2003, when all 13 turbines were operational. Behavioural data were only collected between September and December 2005 from between turbines 8 and 14. Extrapolating from these data to estimate the total number of collisions expected in each year in the absence of any avoidance action gives predictions of 550 collisions in 2001 using option 1 of the Band model, 252 using option 2 and 227 using option 3, and 3,262 collisions in each year between 2002 and 2006 using option 1, 1,497 using option 2 and 1,348 using option 3. Based on these analyses, within-windfarm avoidance rates would have been 0.9615 in 2001, 0.9299 in 2002, 0.9189 in 2003, 0.9284 in 2004, 0.9287 in 2005 and 0.9338 in 2006 using option 1. Using option 2, meso-micro avoidance rates would have been 0.9160, 0.8472, 0.8232, 0.8440, 0.8446 and 0.6990. Using option 3, meso-micro-avoidance rates would have been 0.9067, 0.8302, 0.8037, 0.8268, 0.8273 and 0.6656 respectively.
However, bird activity was only recorded around turbines 8 and 14 in October 2001 and October 2005. If we consider collisions recorded around these turbines in each of these time periods, the predicted number of collisions is 103 herring gulls in October 2001 and 145 black-headed gulls, 90 herring gulls and 260 birds in total during October 2005. The actual number of collisions recorded was 1, 6, 4 and 11 respectively, reflecting avoidance rates of 0.9903, 0.9586, 0.9556 and 0.9577 using option 1 of the Band model, 0.9789, 0.3658, 0.7865 and 0.8077 using option 2 of the Band model and 0.9765, 0.1886, 0.7629 and 0.7865 using option 3 of the Band model.
Assessment of methodology
Fatality data have been collected on a regular basis and following a robust methodology. Corrections have been applied to these data to account for the imperfect detection of corpses due to scavenger behaviour and searcher efficiency.
The observational data that have been collected are extremely limited. Data collection has been restricted to the September to December period in a single year. It is unclear how accurately this reflects bird movements within the windfarm over the rest of the study period. This may have a significant, but unquantifiable impact on the final, derived within-windfarm avoidance rates. In addition, it is unclear whether corrections have been applied to the observational data to account for the imperfect detection of birds.
Using the overall data, it is necessary to make both spatial and temporal extrapolations to estimate the avoidance rates. For this reason, we only use the data collected around turbines 8-14 in October 2001 and 2005 to derive representative avoidance rates.
A6.6 Bouin
Dulac, P. 2008. Evaluation de l'impact du parc eolian de Bouin (Vendee) sur l'avifaune et les chauves-souris. Bilan de 5 anness de suivi. Ligue pour la Protection des Oiseaux, Nantes.
Methods
Visual observations and fatality searches.
Weekly searches were carried out for corpses at the foot of turbines between 2002 and 2006. Searches were restricted to a 100 m 2 box centred on each turbine. To aid searching, each box was divided into a grid with squares of 25 m 2.
Observational data were collected from four points, covering 1 km each. Each month a two hour count was made from each point, with a total of 474 hours of observational data collected from the site as a whole between 2002 and 2006.
Seasons / time of day
Data were collected throughout the year and protocols were designed so that full day was covered.
Species
Black-headed gull (16.23 birds/hr), herring gull (2.26 birds/hr), other gulls (2.09 birds/hr).
Conditions data collected under
The observational protocol was designed to collect data throughout the tidal cycle and in all weather conditions.
Location / habitat
Bouin, Baie de Bourgneuf, France (Coastal)
Turbine / array specification
A single row of eight turbines, each with a hub height of 60 m and a diameter of 80 m. Distance between turbines within each row is not described.
Results
At this site, 30 gulls were recovered from turbine bases over the course of a four year study period. Of these, 28 were black-headed gulls, one was a yellow-legged gull and one was a Mediterranean gull. Using option 1 of the Band model, 584 black-headed gulls and 206 'other' gulls were predicted to collide with the turbines, reflecting avoidance rates of 0.9520 and 0.9903 respectively. For option 2, 483 and 354 birds were predicted to collide respectively, reflecting avoidance rates of 0.9421 and 0.9943. For option 3, the corresponding figures were 237 and 251 birds predicted to collide reflecting avoidance rates of 0.8820 and 0.9920. No collisions were recorded for herring gulls, despite a predicted collision rate of 216 per annum, reflecting a within-windfarm avoidance rate of 1 for options 1, 2 and 3 of the Band model.
Assessment of methodology
Fatality data were collected following a robust protocol, with corrections applied to account for birds lost to scavengers and search efficiency. The intensive nature of these searches, weekly over a four year period, is likely to mean that fatality rates were estimated with a high degree of accuracy.
Observational data were collected over a four year period. However, no corrections were applied to account for imperfect detection. Consequently, bird activity in the area and the derived within-windfarm avoidance rates were likely to have been underestimated.
As activity data were a spatial and temporal match for the period over which collision data were collected, these data were included when estimating representative avoidance rates.
A6.7 Buffalo Ridge
Johnson, G.D., Erickson, W.P., Strickland, M.D., Shepherd, M.F., Shepherd, D.A. 2000. Avian Monitoring Studies at The Buffalo Ridge, Minnesota Wind Resource Area: Results of a 4-year study. Western EcoSystems Technology Inc., Wyoming.
Methods
Visual observations and fatality searches.
Fatality searches were carried out within 126 m x 126 m plots, centred on 61 turbines. Searches were carried out every two weeks and observers covered the area by walking parallel transects separated by a distance of 6 m. This was combined with a series of large bird counts carried out every two weeks for a 0.8 km radius surrounding each of six observation stations. During each survey, two 30 minute observations were made, one in the morning and one in the afternoon. In total 70 hours of survey data were collected over the course of the study period.
Seasons / time of day
Fatality searches were carried out throughout the year. Large bird counts were carried out between 0800 and 1600 h and restricted to the period from 15 March to 15 November.
Species
Herring gull (0.1 birds/hour).
Conditions data collected under
No details given.
Location / habitat
Buffalo Ridge, Minnesota, U.S.A. (terrestrial).
Turbine / array specification
143 turbines arranged in 26 rows with between 100 m and 200 m between each turbine. Each 750 kW turbine had a hub height of 50 m and a diameter of 48 m.
Results
At this site, one herring gull was recovered following collision with turbine. Across the study plots as a whole, the average rate at which herring gulls passed through the windfarm was 0.03 birds per hour, reflecting a total of 625 gull movements within the area over the two year study period. Assuming no avoidance behaviour, the total number of collisions expected would have been 3 using option 1 of the Band model, 5 under option 2 of the Band model and 5 under option 3 of the Band model. The collision rate of 1 bird over the study therefore indicates a within-windfarm avoidance rate of 0.6503 using option 1, 0.8149 using option 2 and 0.7923 using option 3.
Assessment of methodology
The methodology was generally sound with a well-structured search likely to detect all corpses within the study area. Corrections were made for both corpses removed by scavengers and also searcher efficiency. The large bird survey also followed a sound methodology, with corrections applied to account for imperfect detection. However, as observations were limited to 0800 to 1600 h and November to March, it is possible that they failed to detect daily or seasonally important gull movements. This may reflect the fact that raptors were the primary concern at this site.
As it was necessary to extrapolate bird activity data spatially to estimate an avoidance rate, these data have not be included when deriving representative avoidance rates.
A6.8 De Put
Everaert, J. 2008. Effecten van windturbines op de fauna in Vlaanderen: Onderzoeksresultaten, discussie en aanbevelingen. INBO, Brussels
Everaert, J. & Kuikjen, E. 2007. Wind turbines and birds in Flanders (Belgium): Preliminary summary of the mortality research results.
Methods
Visual observations and fatality searches.
Systematic fatality searches were carried out once every 14 days between April 2005 and March 2006. Searches were carried within a circular area, with a radius of 100 m, centred on each turbine. No correction factors were used to account for scavengers or imperfect searcher efficiency.
Observational data describing the number of birds passing within 100 m of the turbine hub were collected between January and February 2006, the period in which the corpses were recovered. The resultant data were used to estimate the total number of birds likely to have passed the turbines over this period. In total 18 hours of survey data were collected.
Seasons / time of day
Observational data were collected throughout the day during January and February 2006.
Species
Black-headed gull and common gull (3,186 during the study period).
Conditions data collected under
No details given.
Location / habitat
De Put, Nieuwkapelle, Belgium (terrestrial).
Turbine / array specification
A row of two turbines, each with a hub height of 75 m and a rotor diameter of 100 m.
Results
In January and February 2006, the corpses of two gulls, one common gull and one black-headed gull, were recovered. Based on the number of birds estimated to have passed through the windfarm during the study period, the combined number of collisions predicted in these two species would be 19 using option 1 and none using options 2 and 3. The two recorded collisions therefore reflect a micro-meso avoidance rate of 0.8928 for common and black-headed gulls using option 1, -9.1051 using option 2 and -11.8383 using option 3.
Assessment of methodology
Whilst fatality searches appear to have been relatively robust and intensive throughout the study period, no corrections were applied to account for the imperfect detection of corpses, either through searcher inefficiency or through loss to scavengers. This may have led to an underestimate of the total number of collision victims.
Details of the methodology used to collect observational data of bird behaviour within the windfarm were sparse. In particular, no details were given of the length of observations used to collect data during the study. There also appears to have been no attempt to account for the imperfect detection of birds, meaning the total number passing through the study area may have been an underestimate. This, in turn would also mean that the final within-windfarm avoidance rate had been underestimated.
As bird activity and collision data have been collected concurrently, these data have been included when deriving representative avoidance rates.
A6.9 Gneizdzewo
Zielinski, P., Bela, G. & Marchlewski, A. 2008. Report on monitoring of the wind farm impact on birds in the vicinity of Gniezdzewo (gmina Puck, pomorskie voivodeship)
Zielinski, P., Bela, G. & Marchlewski, A. 2010. Report on monitoring of the wind farm impact on birds in the vicinity of Gniezdzewo (gmina Puck, pomorskie voivodeship)
Zielinski, P., Bela, G. & Marchlewski, A. 2011. Report on monitoring of the wind farm near Gniezdzewo impact on birds (gmina Puck, pomorskie voivodeship)
Zielinski, P., Bela, G. & Marchlewski, A. 2012. Report on monitoring of the wind farm near Gniezdzewo impact on birds (gmina Puck, pomorskie voivodeship)
Methods
Visual observations and fatality searches.
Collision surveys were carried out in the autumns of 2008, 2010, 2011 and 2012 (September-November). Corpse searches were carried out within 70 m radius of each turbine, on average every 2-3 days.
Over the same periods each year (mid-September - mid-November), activity surveys were carried out within the windfarm. Between 60 and 70 hours of observational data were collected each year, with observation sessions lasting up to 6 hours.
Seasons / time of day
Data were collected throughout the day during the autumn migration period in each year.
Species
Great cormorant (0.17-1.44 birds/hr), gulls (3.88-44.14 birds/hr), little gull (0.23 birds/hr), common gull (0.57 -1.73 birds/hr), black-headed gull (0.51-4.94 birds/hr), herring gull (1.06-5.39 birds/hr).
Conditions data collected under
All conditions
Location / habitat
Gniezdzewo, Poland (terrestrial).
Turbine / array specification
An array of 19 turbines arranged in four rows. Each turbine had a rotor diameter of 80 m and a hub height of 80 m.
Results
In the four autumns over which data have been collected, only a single collision involving a gull was recorded, a black-headed gull during the 2010 field season. No site specific flight height data were available, so it was necessary to use the distributions presented in Johnston et al. (2014a) and option 2 of the Band Model to estimate avoidance rates. In the 2010 field season, 460 black-headed gulls were predicted to have passed through the windfarm, with a predicted collision rate of 0.2 birds. The avoidance rate for black-headed gulls during autumn 2010 would, therefore, have been -3.9524, suggesting that a significant number of birds were attracted to the rotor swept area of the turbine. Using option 3 of the Band model, the collision rate was predicted to be 0.1 birds, reflecting a within-windfarm avoidance rate of -8.9238. However, it should be noted that this collision rate is based on a relatively low number of birds passing through the windfarm and as a result may be unreliable. The unusual nature of this result is confirmed as in three additional years of monitoring, no black-headed gull collisions were recorded, despite often higher levels of flight activity. The avoidance rate for cormorants and all other gull species in all years would have been 100%.
Assessment of methodology
The search for collision victims has been robust, with specially trained dogs used to increase detection. However, no corrections have been applied to account for birds lost to scavengers, potentially meaning the collision rates have been under-estimated.
No correction has been applied to the activity surveys to account for the imperfect detection of birds. As a consequence, the total number of birds passing through the area, and therefore potentially the final avoidance rates, may be under-estimated.
As collision and activity data were collected concurrently, from the windfarm as a whole, throughout the study period, they have been included when deriving representative avoidance rates.
A6.10 Greater Gabbard
RPS. 2011. Galloper Wind farm Project Environmental Statement - Technical Appendices 2: Appendix 4: Greater Gabbard post-construction vantage point surveys, RPS, Glasgow
Methods
Visual observations
Two surveyors collected data from 180˚ arcs to the port and starboard sides of a stationary vessel within Greater Gabbard Offshore Windfarm. Each arc had a radius of 2 km and all birds entering each arc were recorded during snapshot counts taken every 15 seconds. The location of the boat and the viewing area, which covered a total of 15.9 km 2, included seven operational turbines and a total of 36 hours of data were collected during the survey. The flight paths of each bird within the viewing area were noted, as was the proportion of time each bird spent at different heights.
Seasons / time of day
Data were collected between 1 st June 2011 and 28 th July 2011, with each survey lasting four hours.
Species
Northern gannet (0.14 birds/hr), Arctic skua (0.03 birds/hr), lesser black-backed gull (3.69 birds/hr), herring gull (0.11 birds/hr), black-legged kittiwake (1.28 birds/hr).
Conditions data collected under
Conditions were limited to sea-states one and two, to ensure the vessel remained as a stable observation platform.
Location / habitat
Greater Gabbard, UK (offshore).
Turbine / array specification
The survey monitored seven operational turbines, each with a hub height of 77.5 m and a rotor diameter of 107 m.
Results
The predicted number of collisions, in the absence of avoidance behaviour, within the 36 hour study period would have been less than 1 bird from each species. However, no collisions were recorded reflecting an avoidance rate of 1.000 for all species over the course of the study period.
Assessment of methodology
No corrections were applied to account for the imperfect detection of birds during the survey. Consequently, the true level of bird activity within the study area is likely to have been underestimated. Additionally, it was not possible to search for carcasses, meaning inferences about avoidance behaviour can only be drawn from the failure of observers to detect a collision with 36 hours of monitoring. Given the low probability of a collision occurring, and the levels of flight activity recorded, this outcome is unsurprising.
Given the limited data collection during the study period, these data have not been included when deriving representative avoidance rates.
A6.11 Groettocht
Krijgsveld, K.L., Akershoek, K., Schenk, F., Dijk, F. & Dirksen, S. 2009. Collision risk of birds with modern large wind turbines. Ardea 97: 357-366.
Methods
Radar observations and fatality searches.
Fatality searches were carried out within a 100 m radius around each turbine every 2-3 days. Searches were carried out by walking parallel transects, each separated by 4-6 m. Searches were carried out between October and December 2004.
Flight movements were quantified using a 12 kW x-band marine surveillance radar overnight between 1800 and 0700 h on 20 October 2004, 22 November 2004 and 22 December 2004, and the number of radar echoes up to 140 m (the maximum turbine height) were estimated as a measure of flux through the windfarm area.
Seasons / time of day
Resultant data reflect overnight collision rates of birds between October and December 2004.
Species
Key movements recorded included gulls travelling between Lake Ijsselmeer and a nearby roost site around dusk and dawn. However, amongst the five corpses encountered, there were only two gull carcasses, a common gull and a herring gull.
Conditions data collected under
Not specified.
Location / habitat
Agricultural area in the Netherlands.
Turbine / array specification
The array consists of a single line of seven turbines, each separated by 285 m. Turbines have a hub height of 78 m and a rotor diameter of 66 m. However, only the areas under five turbines were searched for carcasses.
Results
The average flux of birds through the area was 370 birds/km/hr, reflecting a movement of 873,534 birds through the study period as a whole. Site specific flight height data were not available for the site, so it was necessary to use the distributions presented in Johnston et al. (2014a) to estimate the proportion of birds at collision risk height, and option 2 of the Band model to estimate predicted collision numbers. In total, the remains of five birds (one herring gull, one common gull, one redwing, two unidentified species) were retrieved. Given that it is not possible to relate the radar tracks to individual species, we calculated the probability of collision based on a bird with the characteristics of first a herring gull, giving a predicted collision rate of 2131 birds over the study period, and an overall avoidance rate of 0.9991 based on option 2 and a collision rate of 1648 birds over the course of the study, with an avoidance rate of 0.9988 using option 3.
Assessment of methodology
The total collision rate may be an underestimate as the initial searching rate of once every three days was lowered to once every two days following the outcome of depredation tests. However, all corpses present were likely to be discovered as only turbines where the surrounding vegetation was low were searched for remains. With the exception of concerns over the depredation rate, the fatality searches were robust.
Flux rates were estimated using x-band radar, with the considerable disadvantage that it cannot be used to estimate the flux rates of different species. As a consequence, using individual species collision rates to estimate an avoidance rate may have led to an inaccurate estimate of the true value. In addition, as a single radar echo may represent multiple birds, there was a considerable risk that the true movement of birds through the area was underestimated and that, therefore, the overall avoidance rate was also underestimated.
As it was necessary to extrapolate activity data both spatially and temporally to estimate the avoidance rates, these data have not been used to derive representative avoidance rates.
A6.12 Haverigg
RPS. 2011. Galloper Wind farm Project Environmental Statement - Technical Appendices 2: Appendix 3: Information on gull flight behaviour at operational wind farms and the estimation of avoidance rates for use in the Band Collision Risk Model, RPS, Glasgow.
Methods
Visual observations
In July and August 42 hours of vantage point surveys were carried out at Haverigg Windfarm following the standard SNH vantage point methodology ( SNH 2010).
Seasons / time of day
Surveys were carried out in July and August.
Species
Gulls (19.90 birds/hr).
Conditions data collected under
Not stated.
Location / habitat
Haverigg Windfarm, Cumbria, UK (terrestrial).
Turbine / array specification
Haverigg Windfarm consists of two groups of four turbines. The first four turbines have a hub height of 45 m and a rotor diameter of 42 m, whilst the remaining four, larger, turbines have a rotor diameter of 52 m.
Results
During 42 hours of vantage point observations, a total of 836 gulls, mostly herring and lesser black-backed gulls were recorded entering the windfarm at a rate of 19.90 birds/hr. However, during the observation periods, no collisions were recorded, reflecting an avoidance rate of 1 over the course of the study period under options 1,2 and 3 of the Band model.
Assessment of methodology
The evidence provided by the survey is limited as no corpse searches were carried out in the area surrounding the windfarm. Whilst 42 hours of survey effort were carried out, no collisions were recorded. However, given the likely rarity of collisions occurring, this is unsurprising. Furthermore, the levels of flight activity within the windfarm are likely to have been underestimated as no correction was made for the imperfect detection of birds.
As insufficient monitoring data have been collected to observe collisions, these data have not been included when deriving representative avoidance rates.
A6.13 Hellrigg
Percival, S. 2012. Hellrigg Wind farm: Goose Refuge Monitoring Report Winter 2011-12, Ecology Consulting, Durham
Percival, S. 2013. Hellrigg Wind farm: Goose Refuge Monitoring Report Winter 2012-13, Ecology Consulting, Durham
Methods
An area covered by a 100 m radius around the base of each turbine was searched between December and March in the winters of 2011/12 and 2012/13 on a weekly basis. Searches were carried out slowly and carefully, with particular care taken over areas containing large clumps of vegetation. The locations of each corpse were carefully noted, and each was left in place to provide information about decay rates and detectability.
Bird activity data were collected through vantage point surveys from a single point following standard SNH guidance. The flight lines of each species were noted and flight altitudes estimated. In total 38 hours of flight observations were collected in this way each winter.
Seasons / time of day
Data were collected between December and March each year, with effort made to cover dawn and dusk movements of birds as well as general daytime movements of birds.
Species
Common gull (8.47 birds/hr in 2011/12 and 507.17 birds/hr in 2012/13), lesser black-backed gull (0.3 birds/hr in 2011/12 and 0.41 birds/hr in 2012/13), herring gull (3.71 birds/hr in 2011/12 and 72.49 birds/hr in 2012/13), great black-backed gull (0.05 birds/hr in 2011/12 and 0.49 birds/hr in 2012/13), black-headed gull (4.79 birds/hr in 2011/12 and 131.48 birds/hr in 2012/13)
Conditions data collected under
Not stated.
Location / habitat
Hellrigg windfarm, onshore.
Turbine / array specification
An array of four turbines with a hub height of 80 m and a rotor diameter of 82 m.
Results
A single collision involving a herring gull was recorded in 2011/12. Based on the passage rate of 3.71 birds/hr, 13 collisions would have been expected in the absence of avoidance behaviour based on option 1 of the Band model, 3 collisions based on option 2 of the Band model and 2 collisions based on option 3 of the Band model. This reflects avoidance rates of 0.9209, 0.6635 and 0.5133 respectively.
Assessment of methodology
Analysis of the length of time corpses remained at the site, suggested that the mean time to disappearance was 22 days, well in excess of the 7 day search intervals. In combination with the systematic and methodical searches carried out at the site, this suggests it is unlikely any corpses went undetected.
Bird activity data were collected following standard SNH vantage point methodology. However, as no correction was made for imperfect detection, the levels of flight activity at the site and, therefore, the overall avoidance rates, may have been underestimated.
As mortality and activity data were collected concurrently at the site, following robust methodologies, these data were used when deriving representative avoidance rates.
A6.14 Keewaunee County
Howe, R.W., Evans, W. & Wolf, A.T. 2002. Effects of Wind Turbines on Birds and Bats in Northeastern Wisconsin. Report to Wisconsin Public Service Corporation and Madison Gas and Electric Company.
Methods
Visual observations and fatality searches.
Intensive searches were carried out between July 1999 and July 2001. Searches were carried out at least once a week. Surveyors visited a 60 m x 60 m area centred on each of the turbines and covered a series of nine 60 m transects in each. These searches were complemented by a series of 3,214 3 minute short counts carried out on 160 dates between 1998 and 2001, to estimate the number of birds within the area.
Seasons / time of day
Surveys were carried out between June and November, with a bias towards data collection during the morning.
Species
Herring gull (0.012 birds/hour), Franklin's gull (0.019 birds/hour), ring-billed gull (1.589 birds/hour).
Conditions data collected under
No details given.
Location / habitat
Keewaunee County, Wisconsin, U.S.A. (terrestrial).
Turbine / array specification
31 turbines with a hub height of 65 m and a rotor diameter of 47 m, within three clusters of 8, 9 and 14 turbines. Distance between turbines within each row is not described.
Results
At this site, one herring gull was recovered following collision with turbine. Across the study region as a whole, the average rate at which herring gulls passed through the area was 0.012 birds per hour, reflecting a total of 131 gull movements within the area over the two year study period. No site specific flight height data were available, meaning it was necessary to use the flight height distributions presented in Johnston et al. (2014a) and option 2 of the Band model. Assuming no avoidance behaviour, no collisions would have been expected under options 2 or 3 of the Band model. The collision rate of 1 bird over the study therefore indicates a within-windfarm avoidance rate of -12.0935 using option 2 and -13.5238 using option 3.
Assessment of methodology
The methodology was generally sound with a well-structured search likely to detect all corpses within the study area. Corrections were made for both corpses removed by scavengers and also searcher efficiency. However, no corrections were made to account for imperfect detectability during the bird surveys.
As it was necessary to extrapolate bird activity data spatially to estimate avoidance rates, these data have not been used to derive representative avoidance rates.
A6.15 Kessingland Windfarm
Wild Frontier Ecology. 2013. Kessingland Wind farm Annual Post-construction Monitoring Report Year 2. Wild Frontier Ecology, Norfolk.
Methods
Visual observations and fatality searches.
Fatality searches were undertaken around the bases of each turbine on nine occasions between November 2012 and March 2013. Surveyors walked a series of transects, separated by 10 m, within 65 m of the turbine base to search for corpses. A corpse correction factor of 1.79 was applied to account for corpses removed by scavengers.
Bird activity was monitored within the windfarm through nine two-hour vantage point surveys at each turbine carried out between November 2012 and March 2013. In total 36 hours of survey effort was completed throughout the study period.
Seasons / time of day
Data collection was carried out over winter 2012/13, between November and March. Surveys were carried out for two hour periods between 0800 and 1500 h.
Species
Black-headed gull (48.5 birds/hr), common gull (15.69 birds/hr), lesser black-backed gull (5.5 birds/hr), herring gull (28.36 birds/hr), great black-backed gull (0.14 birds/hr).
Conditions data collected under
No details given.
Location / habitat
Kessingland, Suffolk, UK (terrestrial).
Turbine / array specification
Two turbines with hub heights of 80 m and rotor diameters of 92 m. Distance between turbines within each row is not described.
Results
Black-headed, common, lesser black-backed, herring and great back-backed gulls were recorded within the study area at varying frequencies. Three gulls were found to have collided with the turbines - one black-headed gull, one common gull and one herring gull. After applying corpse correction factors, these estimates were revised to 1.79 birds of each species. No site specific flight height data were available, so it was necessary to use the modelled flight height distributions presented in Johnston et al. (2014a) and option 2 of the Band model. Given the number of birds likely to have passed through the windfarm during the study period, the predicted collision numbers would have been 28, 21 and 76 respectively. Using option 2, the avoidance rate for black-headed gull would therefore be 0.9367, for common gull it would be 0.9147 and for herring gull it would be 0.9764. Using option 3, the expected collision rates were 13, 12 and 51 respectively, reflecting avoidance rates of 0.8664, 0.8505 and 0.9647. No collisions were recorded involving lesser or great black-backed gulls, reflecting avoidance rates of 1.000 for these species.
Assessment of methodology
The fatality searches appear to have been robust, with corpse correction factors applied to account for loss of corpses to scavengers. However, during vantage point surveys, no corrections were applied to account for imperfect detection. As a result, bird activity within the area was likely to be underestimated, and therefore, the final, derived avoidance rates were also likely to be underestimated.
As collision and bird activity data were collected concurrently over the same area, these data were included when deriving representative avoidance rates.
A6.16 Kleine Pathoweg
Everaert, J. 2008. Effecten van windturbines op de fauna in Vlaanderen: Onderzoeksresultaten, discussie en aanbevelingen. INBO, Brussels
Everaert, J. & Kuikjen, E. 2007. Wind turbines and birds in Flanders (Belgium): Preliminary summary of the mortality research results.
Methods
Throughout 2005 and 2006, an area covered by a 100 m radius around the base of each turbine was searched for collision victims once every 2 weeks. Correction factors were applied to the resultant data to account for searcher efficiency and the removal of corpses by scavengers.
Between September and December 2005, bird activity data were collected between turbines 3 and 7. Data were collected from 2 hours before sunrise to 4 hours after sunset and presented as an average number of birds/day - reflecting an average of 16 hours of survey effort over this period.
Seasons / time of day
Bird activity data were collected between September and December, from 2 hours before sunrise to 4 hours after sunset.
Species
Black-headed gulls (345 birds/day), 'large' gulls (327 birds/day).
Conditions data collected under
Not stated.
Location / habitat
Kleine Pathoweg (Belgium), terrestrial.
Turbine / array specification
A line of 7 turbines, each separated by 280 m. Turbines had a hub height of 85 m and a rotor diameter of 70 m.
Results
In 2005, 240.9 gulls were believed to have collided with turbines once corrections had accounted for imperfect corpse detection. In 2006, this figure was 220.3. Based on a passage rate of 42 birds per hour, in 2005 these figures reflect an avoidance rate of 0.8795 using option 1 of the Band model, -0.2529 using option 2 of the Band model and -0.6887 using option 3 of the Band model. In 2006, these figures reflect an avoidance rate of 0.8898 using option 1 of the Band model, -0.1458 using option 2 of the Band model and -0.5443 using option 3 of the Band model.
Assessment of methodology
Fatality data have been collected on a regular basis and following a robust methodology. Corrections have been applied to these data to account for the imperfect detection of corpses due to scavenger behaviour and searcher efficiency.
The observational data that have been collected are extremely limited. Data collection has been restricted to the September to December period in a single year. It is unclear how accurately this reflects bird movements within the windfarm over the rest of the study period. This may have a significant, but unquantifiable impact on the final, derived within-windfarm avoidance rates. In addition, it is unclear whether corrections have been applied to the observational data to account for the imperfect detection of birds.
As it has been necessary to make spatial and temporal extrapolations to estimate avoidance rates, these data have not been used when deriving representative avoidance rates.
A6.17 Oosterbium
Methods
Visual observations and fatality searches.
Searches were carried out within a 50 m radius of the base of each turbine in autumn 1990 and spring 1991. Searches were carried out on 25 days in the spring and 40 days during autumn. All corpses were assessed in order to determine the cause of death and identify those killed by turbines. Corrections were applied to the data to account for searcher efficiency and scavenger activity.
Bird activity within the windfarm and a surrounding 500 m buffer was assessed during spring 1991 and autumn 1990. These activity levels were used to extrapolate the number of bird-days spent within the windfarm for each species or group of species.
Seasons / time of day
Data covered both the nocturnal and diurnal movements of birds in the spring and autumn.
Species
Gulls (158,600 bird days, autumn 1990; 43,800 bird days, spring 1991).
Conditions data collected under
No details given.
Location / habitat
Oosterbierum, Netherlands (terrestrial)
Turbine / array specification
A cluster of 18 turbines with hub heights of 35 m and a rotor diameter of 30 m, situated within 55 hectares of farmland. Distance between turbines within each row is not described.
Results
Gulls were recorded within the area more commonly during the autumn than the spring. However, the number of collisions was greatest during the spring, when 37 corpses were recovered in comparison to 12 in the autumn. No site specific flight height data were available so it was necessary to use the modelled distributions presented in Johnston et al. (2014a) and option 2 of the Band model. During the autumn, the predicted number of collisions in the absence of avoidance was 883 birds. Therefore, the 12 collisions recorded during the autumn reflects a meso-micro avoidance rate of 0.9864. Using option 3, the predicted number of collisions was 846, reflecting a meso-micro avoidance rate of 0.9858. During the spring, the predicted number of collisions in the absence of avoidance was 244 using option 2 and 234 using option 3. Therefore, the 37 collisions recorded during the spring reflects a meso-micro avoidance rates of 0.8483 and 0.8417 respectively.
Assessment of methodology
Fatality searches were carried out intensively throughout the spring and autumn seasons. They followed a robust methodology with corrections made for both searcher efficiency and scavenger activity.
Activity data were collected throughout the period covered by the fatality searches. However, it appears no corrections were made to the data to account for imperfect detection, meaning activity levels in the area may have been underestimated. As a consequence, the number of collisions predicted in the absence of avoidance, and therefore the derived avoidance rate would also have been underestimated.
As activity and mortality data were collected concurrently and no spatial extrapolation was necessary, these data were used when deriving representative avoidance rates.
A6.18 Waterkapptocht
Krijgsveld, K.L., Akershoek, K., Schenk, F., Dijk, F. & Dirksen, S. 2009. Collision risk of birds with modern large wind turbines. Ardea 97, 357-366.
Methods
Radar observations and fatality searches.
Fatality searches were carried out within a 100 m radius around each turbine every 2-3 days. Searches were carried out by walking parallel transects, each separated by 4-6 m. Searches were carried out between October and December 2004.
Flight movements were quantified using a 12 kW x-band marine surveillance radar overnight between 1800 and 0700 h on 18 October 2004, 17 November 2004 and 20 December 2004, and the number of radar echoes up to 140 m (the maximum turbine height) were estimated as a measure of flux through the windfarm area.
Seasons / time of day
Resultant data reflect overnight collision rates of birds between October and December 2004.
Species
Key movements recorded included gulls travelling between Lake Ijsselmeer and a nearby roost site around dusk and dawn. However, amongst the seven corpses encountered, there was only a single gull carcass, that of a black-headed gull.
Conditions data collected under
Not specified.
Location / habitat
Agricultural area in the Netherlands.
Turbine / array specification
The array consists of a single line of eight turbines, each separated by 300 m, with a larger 1 km gap between turbines 4 and 5. Turbines have a hub height of 78 m and a rotor diameter of 66 m. However, only the areas under five turbines were searched for carcasses.
Results
The average flux of birds through the area was 251 birds/km/hr, reflecting a movement of 1,195,011 birds through the study period as a whole. In total, the remains of seven birds (one common pheasant, one oystercatcher, one black-headed gull, one skylark and two goldcrests) were retrieved. No site specific flight height data were available, so it was necessary to use the modelled distributions presented in Johnston et al. (2014a) and option 2 of the Band model. Given that it was not possible to relate the radar tracks to individual species, we calculated the probability of collision based on a bird with the characteristics of a black-headed gull, giving a predicted collision rate of 1,446 birds over the study period, and an overall avoidance rate of 0.9952. Using option 3, the predicted number of collisions was 1,118 birds, reflecting an overall avoidance rate of 0.9937.
Assessment of methodology
The total collision rate may have been an underestimate as the initial searching rate of once every three days was lowered to once every two days following the outcome of depredation tests. However, all corpses present were likely to be discovered as only turbines where the surrounding vegetation was low were searched for remains. With the exception of concerns over the depredation rate, the fatality searches were robust.
Flux rates were estimated using x-band radar, with the considerable disadvantage that it cannot be used to estimate the flux rates of different species. As a consequence, using individual species collision rates to estimate an avoidance rate may lead to an inaccurate estimate of the true value. In addition, as a single radar echo may represent multiple birds, there was a considerable risk that the true movement of birds through the area was underestimated and that therefore the overall avoidance rate has also been underestimated.
As it was necessary to make temporal and spatial extrapolations with these data, they were not used to derive representative avoidance rates.
A6.19 Yttre Stengrund/Utgrunden Offshore Windfarm
Petterson, J. 2005. The impact of Offshore Wind farms on Bird Life in Southern Kalmar Sound, Sweden: A final report based on studies 1999-2003. Lund University.
Methods
Visual observations
Field data were collected from three observation points located within the Southern Kalmar Sound - Eckelsudde in Oland in the east of the observation area, Olsang in the west of the observation area and Utgrunden Lighthouse in the centre of the Sound of Kalmar. The observation points made it possible to cover the whole of the Sound of Kalmar, including both windfarm sites. The sound was divided into four 5 km zones, each of which was further subdivided into 1-2 km wide zones. The observation point at Olsang covered the first of these 5 km zones, the Utgrunden Lighthouse covered the second and third 5 km zones and the Eckelsudde observation point, the fourth. Observers recorded to the exact minute the location of all flocks of migrating waterbirds they encountered, so that data could be combined into a single dataset at a later date.
Seasons / time of day
Data were collected throughout the spring (22 March to 8 April) and autumn (6 to 28 October) migration periods between 2001 and 2003.
Species
Conditions data collected under
All conditions.
Location / habitat
Southern Kalmar Sound, Sweden (offshore).
Turbine / array specification
Five 2 MW turbines with a hub height of 60 m and a rotor diameter of 72 m at Yttre Stengrund.
Seven 1.5 MW turbines with a hub height of 65 m and a rotor diameter of 70 m at Utgrunden. Distance between turbines within each row is not described.
Results
No collisions were recorded amongst any species during the spring migration periods, reflecting an avoidance rate of 1. No site specific flight height data were available at this site, so it was necessary to use the modelled distributions presented in Johnston et al. (2004). A single collision event was recorded involving four common eider during autumn 2003, reflecting an avoidance rate of 0.1861 using option 2 of the Band model and -0.1098 using option 3. No other collisions were recorded amongst other species, again indicating an avoidance rate of 1.
Assessment of methodology
Methodology is sound with careful calibration of estimates of distance between observers and co-ordination of counts to minimise double-counting. However, there was no correction applied to account for imperfect detection, meaning the total number of birds may have been under-estimated.
As insufficient data have been collected to detect a collision amongst any of the priority species, these data have not been used to derive representative avoidance rates.
A6.20 Zeebrugge
Everaert, J. 2008. Effecten van windturbines op de fauna in Vlaanderen: Onderzoeksresultaten, discussie en aanbevelingen. INBO, Brussels
Everaert, J. & Kuikjen, E. 2007. Wind turbines and birds in Flanders (Belgium): Preliminary summary of the mortality research results. INBO, Brussels
Everaert, J. & Stienen, E.W.M. 2007. Impact of wind turbines on birds in Zeebrugge (Belgium) Significant effect on breeding tern colony due to collisions. Biodiversity and Conservation, 16, 3345-3359
Everaert, J., Devos, K. & Kuijken, E. 2002. Windturbines en vogels in Vlaanderen: Voorlopige onderzoeksresultaten en buitenlandse beviningen. INBO, Brussels
Methods
Visual observations and fatality searches.
Between 2001 and 2007 systematic fatality searches were carried out within a 50 m radius around the base of turbines on a fortnightly basis, increasing to 3-4 times a week during the breeding season. Every turbine was searched, and corrections were made to account for searcher efficiency and scavenger activity.
An initial set of bird activity surveys were carried out at the site in 2000 and 2001. Bird activity within a 400 m section of the breakwater was monitored on four days between June and July in 2000 and 2001, with eight days data collected in total. An additional four days of monitoring were carried out on four days and two nights between September and October 2001.
In June 2004 and 2005, a second set of bird activity were carried out. In each year, two full days of monitoring data were collected covering the period from dawn to dusk. During this period, data were collected between turbines 7 and 12, covering a 720 m section of the breakwater.
Seasons / time of day
Fatality searches were carried out throughout the year. Activity surveys were limited to the breeding season and autumn. Data were collected throughout the day between dawn and dusk, with additional nocturnal surveys carried out during the autumn.
Species
Gulls (234 birds/day), little tern (375-1,860 birds/day), common tern (4,228-10,263 birds/day), Sandwich tern (11-12,334 birds/day).
Conditions data collected under
No details given.
Location / habitat
Zeebrugge, Belgium (Coastal)
Turbine / array specification
25 turbines arranged along Zeebrugge Harbour breakwater. Turbines vary in size from hub heights of 23-55 m and rotor diameters of 22-48 m. Details of collisions at individual turbines are not given, so avoidance rates are estimated assuming turbines with a hub height of 34 m and rotor diameter of 34 m, the most common turbine within the windfarm. Distance between turbines within each row is not described.
Results
Collisions were recorded in every year. For Sandwich terns, collisions varied from seven to 54 birds per year. Using option 1 of the Band model, the estimated number of collisions per year, in the absence of avoidance behaviour, varied from 6,383 birds to 10,299, 8,024 to 10,326 using option 2 and 5,984 to 8,035 using option 3. The meso-micro avoidance rates derived from the values are 1 between 2001 and 2003, 0.9915 in 2004, 0.9972 in 2005, 0.9992 in 2006 and 0.9993 in 2007 using option 1, and 1 between 2001 and 2003, 0.9948 in 2004, 0.9963 in 2005, 0.9989 in 2006 and 0.9991 in 2007 using option 2. Using option 3, the avoidance rates are 1 between 2001 and 2003, 0.9933 in 2004, 0.9952 in 2005, 0.9986 in 2006 and 0.9989 in 2007. Collision data were also obtained relating to June 2004 and June 2005, the periods in which bird activity data were collected and relating to only the turbines around which activity was monitored. In both years, 3 Sandwich terns were observed to have collided between turbines 7-12 in June. Given passage rates of 896 birds/hr in June 2004 and 725 birds/hr in June 2005, this reflects an avoidance rate in 2004 of 0.9895 using option 1 of the Band model, 0.9935 using option 2 of the Band model and 0.9917 using option 3 of the Band model. In 2005, the corresponding values are 0.9940, 0.9920 and 0.9897.
For little terns, collisions varied from two to 12 birds per year. Using option 1 of the Band model, the estimated number of collisions per year, in the absence of avoidance behaviour, varied from 990 birds to 1,087, 165 to 838 using option 2 and 128 to 650 using option 3. The meso-micro avoidance rates derived from the values are 0.9923 in 2001, 0.9914 in 2002, 0.9904 in 2003, 0.9950 in 2004, 0.9982 in 2005, 0.9963 in 2006 and 0.9890 in 2007 using option 1, and 0.9516 in 2001, 0.9455 in 2002, 0.9395 in 2003, 0.9940 in 2004, 0.9884 in 2005, 0.9768 in 2006 and 0.9304 in 2007 using option 2. Using option 3, the avoidance rates were 0.9516 in 2001, 0.9455 in 2002, 0.9395 in 2003, 0.9940 in 2004, 0.9884 in 2005, 0.9768 in 2006 and 0.9304 in 2007. No little tern collisions were recorded in the June 2004 and 2005 data relating to turbines 7-12.
For common terns, collisions varied from 12 to 164 birds per year. Using option 1 of the Band model, the estimated number of collisions per year, in the absence of avoidance behaviour, varied from 4,503 birds to 6,869, 2,475 to 6,530 using option 2 and 1,931 to 5,094 using option 3. The meso-micro avoidance rates derived from the values are 0.9970 in 2001, 0.9977 in 2002, 0.9951 in 2003, 0.9758 in 2004, 0.9812 in 2005, 0.9761 in 2006 and 0.9834 in 2007 using option 1, and 0.9919 in 2001, 0.9939 in 2002, 0.9871 in 2003, 0.9833 in 2004, 0.9501 in 2005, 0.9365 in 2006 and 0.9559 in 2007 using option 2. Using option 3, meso-micro avoidance rates were 0.9896 in 2001, 0.9922 in 2002, 0.9834 in 2003, 0.9786 in 2004, 0.9360 in 2005, 0.9186 in 2006 and 0.9434 in 2007. Collision data were also obtained relating to June 2004 and June 2005, the periods in which bird activity data were collected and relating to only the turbines around which activity was monitored. In 2004 6 common terns were observed to have collided between turbines 7-12 in June, in 2005, this figure was 9. Given passage rates of 603 birds/hr in June 2004 and 248 birds/hr in June 2005, this reflects an avoidance rate in 2004 of 0. 9703 using option 1 of the Band model, 0.9796 using option 2 of the Band model and 0.9738 using option 3 of the Band model. In 2005, the corresponding values are 0.9720, 0.9255 and 0.9045.
For gulls, collisions varied from 110 to 354 birds per year. Using option 1 of the Band model, the estimated number of collisions per year, in the absence of avoidance behaviour, varied from 2,334 birds to 2,537, 2,856 to 3,104 using option 2 and 2,698 to 2,932 using option 3. The meso-micro avoidance rates derived from the values are 0.8979 in 2001, 0.8481 in 2002, 0.8817 in 2003, 0.9105 in 2004, 0.9173 in 2005, 0.9547 in 2006 and 0.9092 in 2007 using option 1, and 0.9166 in 2001, 0.8758 in 2002, 0.9033 in 2003, 0.9268 in 2004, 0.9324 in 2005, 0.9630 in 2006 and 0.9258 in 2007 using option 2. Using option 3 meso-micro avoidance rates were 0.9117 in 2001, 0.8686 in 2002, 0.8976 in 2003, 0.9226 in 2004, 0.9285 in 2005, 0.9608 in 2006 and 0.9214 in 2007.
Data were also obtained relating to black-headed, lesser black-backed and herring gull collisions in June-July 2000, June-July 2001 and September-October 2001, periods corresponding to the times during which gull activity data were collected and restricted to the turbines around which gull data were collected. No collisions were reported involving black-headed gulls. In June-July 2000, a single collision was reported involving a herring gull, reflecting an avoidance rate of 0.9861 using option 1, 0.9829 using option 2 and 0.9819 using option 3. In June-July 2001 and September-October 2001, two collisions were reported involving herring gulls, reflecting avoidance rates of 0.9722 and 0.9976 respectively using option 1, 0.9659 and 0.9959 using option 2 and 0.9639 and 0.9957 using option 3. Single collisions were reported involving lesser black-backed gulls in each of June-July 2001 and September to October 2001, reflecting avoidance rates of 0.9706 and 0.9990 respectively using option 1, 0.9680 and 0.9977 using option 2 and 0.9656 and 0.9975 using option 3.
Assessment of methodology
The study at Zeebrugge offers one of the most comprehensive datasets for collisions involving marine birds. Fatality data have been collected over a seven year period following a robust methodology with corrections made to account for searcher efficiency and scavenger activity. However, a key limiting factor in the dataset is the accompanying bird activity data. In the case of terns, activity data is limited to the period of peak tern activity in June. As a consequence, extrapolating from this to cover the full period when terns are present is likely to vastly over-estimate activity in the area, and therefore the predicted collision numbers. This means that the avoidance rates derived for each year are likely to be significantly over-estimated. This reflects the limitations in the way data are presented within the reports. Ideally, collisions would be broken down on a month by month and turbine by turbine basis, so that avoidance rates could be calculated for the areas in which activity data were collected, rather than extrapolating across the windfarm as a whole.
We used only the collision data collected from gulls during the period in which activity data were collected, and from only those turbines around which activity data were collected, in deriving representative avoidance rates.
Contact
There is a problem
Thanks for your feedback