Scottish Marine and Freshwater Science Volume 5 Number 16:The Avoidance Rates of Collision Between Birds and Offshore Turbines

Appendix 2 Evidence review macro-response - displacement and attraction studies

A2.1 Egmond aan Zee

Leopold, M.F., Dijkman, E.M. & Teal, L. 2011. Local Birds in and around the Offshore Windfarm Egmond aan Zee ( OWEZ) (T-0 & T-1, 2002-2010). Texel, The Netherlands: Wageningen IMARES.

Leopold M.F., Camphuysen C.J., van Lieshout S.M.J., ter Braak C.J.F. & Dijkman E.M. 2004. Baseline studies North Sea windfarms: Lot 5 marine birds in and around the future site Nearshore Windfarm (NSW). Alterra-rapport 1047.

Lindeboom, H.J., Kouwenhoven, H.J., Bergman, M.J.N., Bouma, S., Brasseur, S., Daan, R., Fijn, R.C., de Haan, D., Dirksen, S., van Hal, R., Hille Ris Lambers, R., ter Hofstede, R., Krijgsveld, K.L., Leopold, M. & Scheidat, M. 2011. Short-term ecological effects of an offshore windfarm in the Dutch coastal zone; a compilation. Environmental Research Letters 6. doi:10.1088/1748-9326/6/3/035101.

Location/habitat

Marine 10-18 km offshore.

Turbine /array specification

Hub height 70 m and a rotor diameter 90 m (rotor altitude min 25 m, max rotor altitude 115 above mean sea level). Turbine array consists of 36 Vestas V90 3 MW turbines covering an area 27 km ². Distance within turbines is 650 m within rows and 1000 m between rows.

Methods

The focus of Leopold et al. (2004) and (2011) was to look at avoidance and attraction by birds to the windfarm at Egmond aan Zee for what were termed local birds (although the survey work did cover the Princess Amalia windfarm site, results specific to this windfarm site were not presented). Survey periods covered the pre-construction and post-construction phases of the development. Lindeboom et al. (2011) reported the impacts of the windfarm on a range of taxonomic groups but with respect to birds presented less detail than the above reports and therefore is not considered further here.

The study area was approximately 725 km ² (22 x 33 km). It was selected on the basis that it would include an adjacent offshore windfarm, Princess Amalia, and an anchorage area, where ships wait before entering the nearby major port. Ten transect lines were selected running east to west at distances of 2.47 km apart (with eight additional transect lines added in 2008 running north east to south west). The aim was to cover each transect twice (this was possible until the additional transect lines were added) and the transect lines were sailed in the same order each survey period. Successfully completed surveys ranged between 4-8 days in duration.

Ship based strip census surveys based on the methods adopted in the baseline studies in 2002-2004 (described in Leopold et al. 2004) which were originally derived from Tasker et al. (1984); Komdeur et al. (1992) and Camphuysen and Garthe (2004). All swimming birds were assigned to distance bands: AB (0-100 m); C (100-200 m) and; D (200-300 m) and all observations were assigned to five minute intervals. Flying birds were recorded using the snap shot methodology at intervals of 1 min.

Although BACI design was originally set to look at bird responses to the windfarm, there was considerable annual variation in seabird presence which hampered the ability to look for any differences between pre-construction and post-construction. Therefore the results focussed on comparisons within surveys ( e.g. species-specific monthly counts). Presence/absence data were used as the response in Generalised Additive Mixed Models, which took into account temporal auto-correlation, for all individual species/month combinations there were sufficient data for. Otherwise a more simple General Additive Model was used or, in some cases, statistical models could not be run (birds were counted less than 10 occasions). Therefore, the number of surveys that were available for further analyses varied according to species and were a reflection of the relative abundance of birds each month. Presence /absence data were argued to be more appropriate as they were less affected by the large numbers of zero counts or the few counts with very large numbers of birds recorded. These models took into account the distance to coast, the northing value and the presence of impact area as factors (Egmond aan Zee, Princess Amalia and the anchorage area were considered individually within these models). The model output was then used to predict and subsequently map the probability of birds occurring across the survey area.

Within surveys, there was the possibility of four outcomes: attraction (probability of finding birds inside the windfarm was significantly higher than expected on the basis of the general distribution pattern); avoidance (probability of finding birds inside the windfarm was significantly lower than expected); indifference (probability of finding birds within the perimeter was not impacted by the windfarm and insufficient data.

Study period

Baseline/pre-construction surveys: T-0 = September and October 2002; April, May, June, August and November 2003; February 2004 (described in Leopold et al. 2004).

Post-construction surveys: T-1a = April, June, August, September, November (incomplete) 2007 and January 2008 (May was not repeated); T-1b = April, June, August, September (incomplete), November 2008; January, 2010; T-1c = April, June, August, October (September not possible) November 2009 and; January and February 2010.

Species

Local seabirds as defined as those which reside for some time in the study area. Species accounts were presented for: diver spp, great crested grebe, northern fulmar, northern gannet, great cormorant, common scoter, little gull, black-headed gull, common gull, lesser black-backed gull, herring gull, great black-backed gull, black-legged kittiwake, Sandwich tern, common/arctic tern, common guillemot and razorbill).

Conditions data collected under

Generally aimed to survey in conditions with a Beaufort scale of less than 6 Bft but there were a number of transects that were carried out in higher winds of 6-7 Bft (when light conditions permitted).

Results

Northern gannet: Northern gannet tended to occur on all sides around Egmond aan Zee windfarm but rarely within the perimeter of the windfarm ^[30] . Observations recorded that those few birds that did enter only went one turbine deep. Where presence/absence analyses were possible for the post-construction period (n = 10 surveys), it was shown that the presence of the species was significantly negatively related to the Egmond aan Zee windfarm for only two surveys. Anecdotally it was reported that gannets never entered Princess Amalia Windfarm (which has a higher turbine density ^[31] ). Also highlighted was the lack of searching feeding, resting in the windfarms during the surveys.

Lesser black-backed gull: It was evident that lesser black-backed gulls were often seen within perimeters of windfarm ^[32] . These birds tended to be either resting on the water or foundation structures or feeding at the tidal wakes around the monopiles. Presence/absence analyses for the post-construction period (n = 12 surveys), found that the presence of the species was negatively related to the Egmond aan Zee windfarm for only one survey (the rest were also negative but insignificant). This was counter to what would have been predicted as large fishing vessels only operated outside the windfarm which should have in effect reduced the numbers of birds inside the windfarm (resulting in an apparent avoidance). Most observations of lesser black-backed birds were anecdotally reported to be associated with, looking out for or resting in the wake of active fishing vessels.

Herring gull: Birds did occur in the windfarm area but overall fewer birds were recorded in the offshore environment compared to other gulls (notably in August where herring gulls remain mostly near shore). Like lesser black-backed gulls they were often associated with fishing vessels. Presence/absence analyses for the post-construction period (n = 14 surveys), found that the presence of the species was negatively related to Egmond aan Zee windfarm for eight surveys although this effect was only significant in three cases. Herring gull distribution patterns were thought to be likely to be attributable to overall latitudinal variation, as evidenced by the strong effect of distance to coast in the models (significant p values for six surveys).

Great black backed gull: Birds were reported as occurring in the windfarm area ^[33] . Presence/absence analyses for the post-construction period (n = 18 surveys), found that the presence of the species was positively related to the Egmond aan Zee windfarm in five cases, four significantly, although this effect was only apparent at low densities. There were also two surveys in which significant effects were reported. As reported for lesser black-backed gull, birds did tend to feed around fishing vessels but not in the same high numbers.

Black-legged kittiwake: birds were recorded within the windfarm and in general numbers declined with decreasing distance to shore (apart from in November and one January). Presence/absence analyses for the post-construction period (n = 5 surveys), found that the presence of the species was positively related to the Egmond aan Zee windfarm in three cases, one significantly.

Assessment of methodology

Overall, there was lack of consistent evidence for either displacement or attraction for any of the species. This could have been partly due to the importance of factors operating at the larger scale of study area. For the larger gull species, there was a strong association with fishing vessels in the study area. Since fishing was no longer permitted in the windfarm areas, this could have confounded any results reported to do with possible attraction or avoidance of windfarms. There was also evidence that distance to coast was an important factor in determining the overall distribution patterns of herring gulls.

There were potential issues relating to the choice of statistical approach. As comparisons of pre-construction and post-construction data was deemed not to be possible, multiple tests for individual surveys were carried out which may have led to the possibility of a Type 2 error (increased chances of reporting a false significant result). Also the numbers of observations were low for northern gannet and gull spp and consequently the modelling power was very low (Lindeboom et al. 2011). Moreover, the model outputs were in the form of p values and model co-efficients which could not be converted into avoidance rates without further details being presented (even if consistent effects had been observed). Therefore, from the results provided, it is not possible to derive displacement/attraction rates or thus macro-response rates for the study species.

A2.2 Robin Rigg

Natural Power. 2014. Analysis of Marine Ecology Monitoring Plan Data from the Robin Rigg Offshore Windfarm, Scotland (Post-construction Year 3). Draft Technical Report. E.ON Climate & Renewables.

Location/habitat

Marine, offshore < 11 km

Turbine /array specification

Turbine array consists of 60 3.0 MW Vestas turbines which are positioned approximately 500 m apart. Turbine specifications are given as turbine towers 80 m high and a rotor blade length of 44 m.

Methods

The purpose of this report was to look at: displacement of key species; changes in patterns of abundance and distribution; compare observed patterns with predicted impacts/sensitivities from the EIA process.

Data collection was carried out during the pre-, during and post-construction periods.

Boat based surveys based on standard European Seabirds-At-Sea ( ESAS) survey methods were carried our ( e.g. prior to the publication of Camphuysen et al. 2004) as used in the baseline period. In order to ensure comparability between the different phases of the development, methods were kept the same throughout. Additional survey work has been carried out from year 3 of the post-construction period which corresponds to current best practice. The main difference between the two approaches is that for flying birds the former records flying birds using transect methodology whereas the latter uses the snap shot methodology currently regarded as best practice. A total of 10 parallel transects running in a south west to north east direction of 18 km in length and spaced 2 km apart.

For the purpose of analyses, each survey was divided into individual blocks of 600 m ² (corresponding to the 300 m either side of the transect line as both sides of the boat are surveyed). In terms of the data, there was a cleaning process applied. Uneven sampling effort across the different phases of the development (some months were surveyed twice) was identified as an issue and therefore a single survey at random was selected. The study area was also cropped to remove an area in the northeast where shallow waters sometimes prevented access and two transects in the southeast were removed due to under surveying during the pre- and during construction phases. There was also a gap during the construction period where there was no building activity (January and July 2008) and these were also excluded from the analysis.

Birds on the water and birds in flight were analysed separately. Datasets that had fewer than 300 non-zero observations were not considered. Raw observations were mapped and summary statistics for the three development phases were calculated in order to provide an initial indication of any change. These included: mean number of sightings (groups of animals), mean number of individuals per segment and mean number of individuals per segment per month. These are not discussed here however and the results of models output are focussed upon.

Distance Sampling techniques were not used to correct the survey counts and a correction factor derived using the detection function was applied instead. Generalised Additive mixed effects mixture modelling carried out within a Bayesian framework were applied in order to deal with zero inflation (high number of zeros). Transect and survey were incorporated as random effects in order to deal with spatial and temporal autocorrelation. Covariates used in the models were latitude, longitude month (or season) and time of day.

Outputs of the models were used to produce density surface maps of the predicted distribution during the three different phases of the development. Abundance and density estimates for each species within the windfarm and the study area were produced for each phase. In order to look at avoidance, model outputs were used to predict the number of animals within the windfarm and for buffers 0.5, 1, 1.5 and 2 km of the three different windfarm phases. Model outputs were presented only for the comparisons of pre-construction to construction and pre-construction to post-construction (but it was not clear which of the spatial scales they related to).

Study period

Baseline surveys: monthly basis between May 2001 and April 2002. Further pre-construction surveys April and May 2003 and then on a monthly basis between January 2004 and September 2004 (excluding April and June) with further work in July 2007. Construction surveys: monthly basis between January 2008 and February 2010 (excluding November 2009). Post-construction: monthly surveys from March 2010 to February 2013 - scheduled to continue until 2015.

Species

Data were collected for a wide range of species ( e.g. seabirds, seaducks, waders, passerines). Species accounts were only presented for the following key species: scaup, common scoter, red-throated diver, Manx shearwater, northern gannet, great cormorant, black-legged kittiwake, herring gull, great black-backed gull, common guillemot, and razorbill.

Conditions data collected under

Not specified but ESAS provide guidance regarding suitability of conditions.

Results

Northern gannet: Modelling of the numbers of northern gannet on the water was not possible as there were too few sightings. The predicted numbers of northern gannet in flight across the three different phases of the development were found not to be significantly different. There appeared to have also been relatively little change in the predicted densities for the windfarm site, windfarm plus buffers (at any of the scales) or even at the level of the study area ^[34] . Although northern gannet was recorded throughout the study area, densities of the gannets were reported as being generally low ^[35] .

Black-legged kittiwake: The predicted numbers of black-legged kittiwake on the water across the three different phases of the development were found not to be significantly different. There appeared to have also been relatively little change in the predicted densities for the windfarm site, windfarm plus buffers (at any of the scales) or even at the level of the study area ^[36] . A similar result was found for black-legged kittiwakes in flight ^[37] .

Herring gull: Modelling of the numbers of herring gull on the water was not possible as there were too few sightings. The predicted number of herring gull in flight across the three different phases of the development were found to be significantly different with the numbers within the windfarm decreasing over the development (pre-construction to construction p = 0.0021, parameter estimate -0.750 and pre-construction to post-construction p = 0.0013, parameter estimate - 0.841).

Great black-backed gull: Modelling of the numbers of herring gull on the water was not possible as there were too few sightings. The predicted number of herring gull in flight were found to significantly differ from pre-construction to construction (p = 0.0166, parameter estimate -1.133) but not from pre-construction to post construction (p = 0.7854).

Assessment of methodology

There were insufficient data to allow modelling of the observations of birds on the water for northern gannet, herring gull, and great black-backed gull. For birds in flight, there was evidence for a significant decrease for herring gull both during the construction and post-construction periods whereas this decrease was only noted during construction for great black-backed gull. Northern gannet and black-legged kittiwake did not appear to respond to the presence of the windfarm. From the results provided, it was not possible to derive macro-response rates since it was not clear what models have been fitted and it was not apparent whether the changes were due to the presence of the windfarm or as result of changes at the scale of the overall study site. It is acknowledged though that despite this being year 3 of the post construction, it is not the final report and any reported results should be considered as preliminary findings.

A2.3 Blighbank

Vanermen, N., Stienen, E.W.M., Courtens, W., Onkelinx, T., Van de walle, M. & Verstraete, H. 2013. Bird monitoring at offshore windfarms in the Belgian part of the North Sea - Assessing seabird displacement effects. Rapporten van het Instituut voor Natuur- en Bosonderzoek 2013 ( INBO.R.2013.755887). Instituut voor Natuur- en Bosonderzoek, Brussel.

Location/habitat

Marine, 42 km offshore

Turbine /array specification

55 turbines. Additional information was not presented.

Methods

This report looked at Blighbank and Thorntonbank windfarms (but also referred to the more recent development of Lodewijckbak) in what is termed the windfarm concession zone located in the north eastern edge of the Belgian Part of the North Sea ( BPNS). Surveys at both windfarms are still on going.

Data collection was carried out during the pre-construction, during and post-construction periods.

A BACI approach was adopted in order to monitor sea bird displacement. A control area of comparable size was selected on the basis of having similar attributes in terms of number of birds, environmental conditions and having sufficient historic data. A buffer zone of 3 km was applied to the boundary of the windfarm (and the control area), in order to reflect the distance to which the effects of the windfarm could be an issue for birds.

Boat transects were carried out on a monthly basis (citing Tasker et al. 1984) from 2008. The time interval used in this survey for recording was 10 minutes (a number of other windfarm surveys use 1 min). Although only transect routes used post 2012 were shown ^[38] , despite some apparent minor shifts in the location the overall configuration was considered to be the same over the whole monitoring period (Nicolas Vanermen pers. comm.). An overview was provided of all the ESAS counts carried out by INBO during the period of 1992-2012 based on location of counts, this could not be used to look at survey effort which varied over the study period ^[39] . Count effort for Blighbank ^[40] (as shown by the number of surveys) indicated overall higher effort in the pre-construction period (but this included data possibly dating back to 1992). There was also marked monthly variation in effort in the preconstruction phase with peaks in February/August for the pre-construction period and in March/December for the post construction period.

Although distance sampling was used to correct count data to estimate the total numbers of birds within the BPNS (based on Buckland et al. 2001), it was not applied for modelling of the windfarm data (this was on the grounds that the correction factor used for both control and the windfarm area was likely to be the same Nicolas Vanermen pers. comm.). In order to analyse the count data, generalised linear models were used, with a negative binomial distribution assumed in order to cope with over dispersion. Modelling was carried out using area (the reference area or the impact area) and month (as a as a continuous variable in order to model seasonality) included as explanatory terms in what was termed the reference model (based on data collected prior to April 2008). The best model was then selected using a backward approach using a Wald test and looking at the resulting AIC values. The impact model was a simple extension of the count component of the reference model with before and after being added as factor variables to the model. Although not carried out in this report, the natural exponent of the model coefficients can be used to derive the factorial change (and hence the overall percentage change in numbers from pre to post construction - see Table A6.1).

Species' preference for the windfarm area was calculated using Jacob's Selectivity Index (calculated using the proportion of birds that occur inside the entire windfarm concession zone compared to the total numbers within the BPNS and the proportion of the surface area of the concession zone to the total area of the BPNS) whereby values of -1 represent total avoidance and + 1 is total preference (attraction). However this data was only carried out for the baseline data and hence are not considered further here.

The impact of the windfarm was considered separately for the post-construction phase at the scale of the windfarm, the windfarm and buffer, and the buffer without the windfarm ^[41] . Displacement-related coefficients and their respective p values were reported.

Study period

The baseline period (reference period) referred to data pre-September 2009. The construction period ran from September 2009 to August 2010, and the post-construction period was from September 2010 onwards. Data collected during the initial construction period were not used in subsequent assessment due to access issues over this period. Results are presented for up until December 2012.

Species

Northern fulmar, northern gannet, great skua, little gull, common gull, lesser black-backed gull, herring gull, great black-backed gull, black-legged kittiwake, common guillemot, and razorbill.

Conditions data collected under

Not specified in the report. Conditions were, however, mostly favourable - boat surveys are cancelled when wave heights > 1.8 m, and in poor visibility (Nicolas Vanermen pers. comm.).

Results

Northern gannet: Model coefficients were significant for the scale of the windfarm and buffer and buffer without the windfarm (see Table A6.1). Therefore there were highly significant decreases in numbers of northern gannet in the windfarm and the buffer of 3 km at all three spatial scales considered.

Lesser black-backed gull: Model coefficients were significant for the windfarm and buffer, and buffer without the windfarm, and were only just not significant for just the windfarm. Therefore there was a significant increase in numbers of lesser black-backed gull in the windfarm and the buffer of 3 km relative to the pre-construction period.

Herring gull: The model coefficient was only significant at the scale of the windfarm, indicating an increase in numbers in the windfarm area relative to the pre-construction period.

Great black-backed gull: The model coefficients were not significant, indicating no changes in numbers of the species relative to the pre-construction period.

Black-legged kittiwake: The model coefficients were not significant, indicating no changes in numbers of the species relative to the pre-construction period.

Table A2.1 Model outputs of Negative binomial modelling converted into factorial changes

Species	Scale	Model coefficient	P value	Factorial Change*	Overall change as a proportion
Northern gannet	Windfarm	-1.83	0.000	0.16	0.84
	Windfarm plus buffer	-1.52	0.000	0.22	0.78
	Buffer	-1.32	0.003	0.27	0.73
Lesser black- backed gull	Windfarm	1.57	0.059	4.81	-3.81
	Windfarm plus buffer	2.39	0.004	10.91	-9.91
	Buffer	2.37	0.006	10.70	-9.70
Herring gull	Windfarm	3.97	0.000	52.98	-51.98
	Windfarm plus buffer	1.26	0.111	-	-
	Buffer	0.83	0.269	-	-
Great black- backed gull	Windfarm	1.08	0.127	-	-
	Windfarm plus buffer	0.47	0.447	-	-
	Buffer	0.54	0.428	-	-
Black-legged kittiwake	Windfarm	0.25	0.605	-	-
	Windfarm plus buffer	0.50	0.264	-	-
	Buffer	0.77	0.092	-	-

*natural exponent of the model co-efficient.

Assessment of methodology

The results of this report should be considered as being preliminary since further data was collected for 2013. Nevertheless, northern gannet was shown to decrease in response to the presence of windfarm by a value of 0.84. This value could be taken as being indicative of macro-avoidance. Whereas both lesser black-backed gull and herring gull shown quite marked attraction to the windfarm. Great black-backed gull and black-legged kittiwake showed no overall response to the windfarm. From the results provided it was not possible to look at seasonal variation in displacement or attraction.

Sampling effort was biased towards the pre-construction phase and was characterised by variable effort on a monthly basis. Spatial coverage over the whole study period is likely to have been fairly consistent however. The data presented in this report is based on a BACI approach and potentially has limited value in looking at changes in the wider area but long term monitoring in the BPNS has continued throughout the study period and hence there is scope to include this at a later stage if required.

A2.4 Thorntonbank

Vanermen, N., Stienen, E.W.M., Courtens, W., Onkelinx, T., Van de walle, M. & Verstraete, H. 2013. Bird monitoring at offshore windfarms in the Belgian part of the North Sea - Assessing seabird displacement effects. Rapporten van het Instituut voor Natuur - en Bosonderzoek 2013 ( INBO.R.2013.755887). Instituut voor Natuur- en Bosonderzoek, Brussel.

Location/habitat

Marine, 27 km offshore.

Turbine /array specification

Initially six turbines, final array to consist of 54 turbines.

Methods

See Appendix 2, section A2.3 for overall approach.

The impact of the windfarm was considered separately for the two different operation phases: phase 1 (turbine array consisting of six turbines) and; phase 2 (second construction period). Models were run at the scale of the windfarm and buffer only ^[42] .

Power analyses were also carried out for the reference data collected in the Thorntonbank study area in order to determine the power required to detect change in numbers of birds (25, 50 and 75% decrease) and the length of the monitoring period required.

Study period

Monthly surveys were started in 2005 (although additional data were available from 1993 based on surveys that have been carried out of the whole region of the BPNS but coverage was uneven spatially and temporally). The baseline period (reference period) referred to data pre-April 2008. The construction period ran from April 2008 to May 2009, and the post-construction period (called here the impact period) was from June 2009 to April 2011. Thereafter there was another period of construction from May 2011 that was ongoing at the time of the report.

Species

Northern fulmar, northern gannet, great skua, little gull, common gull, lesser black-backed gull, herring gull, great black-backed gull, black-legged kittiwake, Sandwich tern, common tern, common guillemot, razorbill.

Conditions data collected under

Not specified but ESAS provide guidance regarding suitability of conditions.

Results

Northern gannet: For both phase 1 and phase 2, the model coefficients were not significant, indicating no changes in numbers of the species relative to the pre-construction period.

Lesser black-backed gull: For phase 1, the model co-efficient was not significant. For phase 2, a significant model co-efficient of 2.13 was reported (p = 0.052) for the scale of the windfarm, indicating a decrease inside the windfarm (but this effect was not found for the other models at the scales of the windfarm plus buffer, and buffer without the windfarm).

Herring gull: For both phase 1 and phase 2, the model coefficients were not significant, indicating no changes in numbers of the species relative to the pre-construction period.

Great black-backed gull: For phase 1, the model co-efficient was reported as 1.5 and was found to be significant (p = 0.024) for the windfarm plus buffer indicating an attraction to the windfarm. Whereas for phase 2, the model coefficients were not significant, indicating no change in numbers of the species relative to the pre-construction period.

Black-legged kittiwake: For both phase 1 and phase 2, the model coefficients were not significant, indicating no changes in numbers of the species relative to the pre-construction period.

Assessment of methodology

The results of this study were derived from when the windfarm only consisted of 6 turbines (phase 1) or during the next phase of construction of a further 48 turbines (phase 2). Hence the years covered by this report do not include the post-construction phase of a fully post-construction windfarm. Hence the results are not considered further here as part of this review.

A2.5 Nysted

Petersen, I.K., Christensen, T.K., Kahlert, J., Desholm, M. & Fox, A.D. 2006. Final results of bird studies at the offshore windfarms at Nysted and Horns Rev, Denmark. Commissioned by DONG Energy and Vattenfall A/S. National Environmental Research Institute.

Location/habitat

See under section 5.4.1.3 under barrier effects.

Turbine /array specification

See under section 5.4.1.3 under barrier effects.

Methods

Aerial transect surveys were carried out using methodology described in Kahlert et al. 2004 (which prior to the publication of Camphuysen et al. 2004 was commonly cited by other studies as the standard methodology). A total of 26 parallel transects running north to south separated by distances of 2 km were carried out covering an area of 1,700 km ². The area was extended by four additional transect lines in 2002 to increase the area to 1,846 km ².

Jacob's selectivity indexes (D) were used in order to look at displacement and attraction. This approach essentially determines bird preferences for the windfarm area and a buffer zone (2 and 4 km) where birds could still be impacted, in relation to their preference to the whole study area. Values fell between -1 (displacement) and +1 (attraction). Bird encounters (for both individuals and groups here termed as clusters) rather than estimates of bird densities were used. Bird preferences were then compared by looking at the pre- and post-construction D values, based on a simple comparison of number rather by formal statistical analyses, in order to describe the change in bird utilisation of the windfarm.

Bird encounter rate (number of birds reported per km of survey route per observer) was used as a proxy of density in order to calculate mean densities in the windfarm area and in the buffer zone. Comparisons of the mean densities pre- and post-construction were carried out using Student's t-test with corrections for unequal variance. Sufficient data (with respect to the five priority species) was available for comparisons for herring gull at Nysted in January and Horns Rev in March.

Study period

Pre-construction period = August 1999 to August 2002 (n = 21 surveys); construction period = January 2003 to August 2003 (n = 3); post-construction period = January 2003 (sic) to November 2005 (n= 8). The timing of the actual surveys ( e.g. by month were not reported). Only the pre-construction and post-construction surveys were used. There was a lack of autumn surveys for the post-construction phase and therefore only winter and spring surveys were available.

Species

Diver spp, great cormorant, long-tailed duck, common eider, common scoter, red-breasted merganser, herring gull and great black-backed gull.

Conditions data collected under

Not specified.

Results

Herring gull: Comparisons of pre- and post-construction selectivity indices for numbers clusters of birds showed no change ^[43] . Whereas selectivity indices for numbers individuals showed a tendency towards decreased selectivity ( e.g. less birds were using the area) for the windfarm as well as both buffer zones ^[44] . There was no significant difference between bird encounter rate between the pre- and post-construction phases in the windfarm area or the 4 km zone but a significant difference was found for the 2 km buffer. The report concluded there was no evidence for either attraction or avoidance.

Great black-backed gull: outputs of the models were all found to be insignificant apart for the selectivity indices for individual birds post-construction and hence are not reported further here as they have no meaningful comparison for pre-construction.

Assessment of methodology

Overall there was little evidence that herring gull showed any response to the presence of the windfarm.

There are a number of potential limitations of the approach used. There may be issues to do temporal coverage - from the information provided, it was difficult to be able to evaluate how sampling effort varied over the different phases of the development. Also whilst the Jacob's selectivity indices may provide an indication of the likely direction of response, these cannot be directly translated into displacement rates. Also the comparison of pre- and post-construction bird encounter rate had limited value since they provided no indication of changes in distribution that may have occurred at a wider scale (and therefore nothing to do with the presence of the windfarm).

A2.6 Horns Rev

Location/habitat

See under Appendix 1, section A1.2.

Turbine /array specification

See under section Appendix 1, section A1.2.

Methods

Aerial surveys: Aerial transect surveys were carried out using methodology described in Kahlert et al. (2004) which prior to the publication of Camphuysen et al. (2004) was commonly cited by other studies as the standard methodology. A total of 26 parallel transects separated by distances of 2 km were carried out covering an area of 1,350 km ².

Study period

Pre-construction period = August 1999 to January 2002 (n = 16 surveys); construction period = March 2002 to August 2002 (n = 3); post-construction period January 2003 to November 2005 (n= 15). The timing of the actual surveys ( e.g. by month were not reported). Only the pre-construction and post-construction surveys were used.

Species

Diver spp, northern gannet, common eider, common scoter, little gull, Arctic/common tern and guillemot.

Conditions data collected under

Not specified.

Results

Northern gannet: There were no observations of northern gannet inside the windfarm pre- or post-construction. Comparisons of pre- and post-construction selectivity indices for the buffer zones indicated increased avoidance at the 2 and 4 km zone. Insufficient numbers of birds were recorded in order to be able look at encounter rates and limited further interpretation of what the likely overall response of northern gannet to the windfarm.

Herring gull: Comparisons of pre- and post-construction selectivity indices for clusters and individuals of birds indicated a reduced avoidance of the windfarm area. The bird encounter rate revealed no significant difference between the pre- and post-construction period. It was concluded that despite an increased preference being found during construction (citing Christensen et al. 2003), attraction was not observed post-construction.

Black-legged kittiwake: Model outputs were not significant for numbers of clusters of birds post-construction and for both pre- and post-construction for numbers of individual birds. Hence the results are not reported here.

Assessment of methodology

See Appendix 2, section A2.5.

A2.7 Alpha ventus demonstration site

Bundesamt fur Seeschiffart und Hydrographie, BSH 2011. Okologische Begleitforshung bei alha ventus erste Ergebnisse (Environmental research at alpha ventus - first results). Contributions from the Event of 10 May 2010, Katholische Akademie Hamburg.

Mendel, B., Kotzerka, J., Sommerfeld, J., Schwemmer, H., Sonntag, N. & Garthe, S. 2014. Effects of the Alpha Ventus offshore test site on distribution patterns, behaviour and flight heights of seabirds. In Ecological Research at the Offshore Windfarm Alpha Ventus: Challenges, Results and Perspectives. Editors Federal Maritime and Hydrographic Agency, Federal Ministry for the Environment, Nature Conservation and Nuclear Safety. Springer Spektrum.

All the post-consent monitoring reports from this OWF demonstration site are written in German (Stefan Garthe pers. comm.). The first reference reviewed is a report ( BSH 2014) which has a full English translation. The second reference (Mendel et al. 2014) is a book chapter and is written in English. Neither reference can be considered to be fully comprehensive in the level of detail provided but given the importance of this OWF site this information should be included. The information which is cited below is largely taken from Mendel et al. (2014).

Location/habitat

45 km offshore

Turbine /array specification

Twelve turbines. Two designs (jacket foundation and tripod steel foundations) - no further information provided.

Methods

Two study areas were selected: the key study area, the size of which was in excess of 30 times the size of the windfarm itself and; a reference site which appeared to be nearly twice the size of the study area. Boat based surveys were carried out according to standard European Seabirds-At-Sea ( ESAS) survey methods. Aerial-based methods were based on methods described in Pihl and Frikke (2002), Noer et al. (2000) and Diederichs et al. (2002) (full citations are given in Mendel et al. 2014). As well as data from the EIA studies, additional data from eight multiple-day ship-based surveys and 21 aerial surveys carried out in both study areas were available. No further information was provided, however ( e.g. on the timing of the surveys in relation to season).

In order to carry out analyses of the changes in distribution patterns for pre-and post-construction data, data were collated into grid cells of 1 km ² and only data from the key study area were used. A total of six species or species groups were looked at (divers, northern gannet, lesser-black backed gull, little gull, black-legged kittiwake and common guillemot) and only the most important period/s for each of these were focussed upon. Data were also collated over large time periods (usually seasons).

Changes in abundance were looked at using the pre- and post-construction data and only two species were considered (lesser-black backed gull and common guillemot). Generalised Linear Mixed Models of the abundances of birds at different distances in relation to the windfarm (0-2 km, 2-6 km and 6-10 km) were tested in three different models using a Poisson error distribution.

The percentage of birds recorded in each behavioural category was calculated for the key study areas and the reference area for lesser black-backed gull only.

Study period

Data from 2000-2008 were regarded as pre-construction (construction started in September 2008) and data from 2010-2012 represented the post-construction period.

Species

Northern gannet, northern fulmar, black scoter, skua spp, gull spp, and auks spp.

Key species: Red-throated diver, black-throated diver, lesser black-backed gull, black-legged kittiwake, little gull, common guillemot and razorbill.

Conditions data collected under

Data collected according to ESAS methods (sea state < 5Bft).

Results

Changes in distribution

The statistical significance of the following results was not provided and interpretation of results was largely based on maps representing densities of birds for the 1 km ² grid cell system of the key study area. Overall lower abundances were reported post-construction for six of the species/groups but only the relevant species are reported further here.

Northern gannet: the impact of the windfarm was hard to qualify due to the very low numbers recorded within the key study area. This species was reported to have occurred on seven occasions (nine individuals) within the windfarm area during the pre-construction period and none were observed post-construction. Data were taken from March to September and hence represented the breeding season.

Lesser-black-backed gull: a 'clear decrease' was reported to have occurred from the pre- to the post-construction period. Although low to medium densities were reported post construction within the windfarm area, the highest densities were found a few kilometres away from the windfarm site (previously some of the highest were found within the perimeter of the windfarm area during pre-construction). Data were taken from May to July and hence represented the breeding season.

Black-legged kittiwake: a 'remarkable decline' occurred post-construction not only within the perimeter of the windfarm but at the scale of the whole key study sites. Numbers recorded overall were very low however ( e.g. highest number of birds recorded per km ² was 5). Data were taken from November to April and hence represented the non-breeding season.

Changes in abundance

Lesser-black backed gull: Statistically significantly lower abundances were reported for the 0-2 km, 2-6 km and 6-10 km distance class and the models suggested that the disturbance effect was strongest within 2 km of the windfarm,

Assessment of methodology

Based on the information provided, it is not possible to carry out a proper assessment of the methodology used. The overall abundance of northern gannet was very low and therefore this study cannot be cited as evidence of the windfarm having an impact on their distribution. There is some evidence to suggest that displacement may be occurring for lesser black-backed gull and black-legged kittiwake based on the maps of the distribution of bird densities for pre- and post- construction, but there was a lack of statistical analyses. However a statistically significant reduction in the abundance of lesser black-backed gulls was reported for all the three distances classes from the windfarm.