Collision assessment for migratory species

404. In order to assess potential collision risk for migratory water birds and seabirds on passage, Scoping Opinion advice was to assess these species with reference to site-specific survey results and the Marine Scotland commissioned update to the 2014 report on ‘strategic assessment of collision risk of Scottish offshore wind farms to migrating birds’ (WWT, 2014).
405. In the absence of the revised update, Scoping Opinion advice was to assess any SPA migratory waterbird species relevant to the Proposed Development which are not considered in the 2014 Report on a qualitative basis. As of August 2022, the updated report was not publicly available, therefore the collision assessment for migratory species was conducted based on the WWT (2014) report, with any SPA migratory waterbird species relevant to the Proposed Development which are not considered in the 2014 Report being assessed on a qualitative basis.
406. The collision assessment for migratory species is presented in paragraph 637 onwards.

Reference Populations

407. For each of the eight key species assessed for collision impacts during the operation phase, relevant reference populations were required for comparison with the number of birds considered likely to suffer mortality in the different bio-seasons across a year. For the breeding season assessment, the total number of breeding adults from all colonies within mean maximum foraging range + 1 S.D. were used, as estimated by Woodward et al., (2019) ( Table 11.9   Open ▸ ).
408. Corresponding reference populations for the BDMPS bio-seasons that make up the non-breeding season were taken from Furness (2015) ( Table 11.9   Open ▸ ).

Parameters used in CRM Assessment

Wind turbine parameters

409. Details of all wind turbine parameters used in the CRM are presented in volume 3, appendix 11.3.

Seabird Densities

410. Monthly densities of flying birds in the Proposed Development only (excluding the 16 km buffer of the Offshore Ornithology study area) were estimated using design-based strip transect methods from the HiDef digital aerial surveys conducted between March 2019 – April 2021. The estimates for all species were based on counts that had been apportioned for non-identified birds during the surveys. Further detail is provided in volume 3, appendix 11.1.
411. Estimates of mean (Developer Approach) and maximum (Scoping Approach) monthly densities and pooled standard deviations (the latter only required for sCRM) for flying birds only were used as input to the CRMs. Further details are presented in volume 3, appendix 11.3.

Seabird Biological Parameters

412. Discussions through the Ornithology Road Map process (Road Map Meeting 3 28 September 2021 and NatureScot advice 7 October 2021) were used to agree sources of seabird morphological and behavioural parameters (for example flight speed and wing span) to parameterise the CRMs. Body length, wingspan and flight speed measurements were sourced from Robinson (2005), Pennycuick (1997) and Alerstam et al. (2007). This information was not available for Arctic tern, so the morphological and behavioural parameters for common tern were used instead as the two species are very similar.
413. NatureScot provided advice for gannet based on an analysis of nocturnal activity of tagged birds which showed there to be very low levels of activity after dark (Furness et al., 2018 and references therein). For herring, lesser black-backed and little gulls, Arctic and common terns and great skua, the nocturnal activity scores were taken from Garthe and Hüppop (2004). The nocturnal activity score for kittiwake was taken from the previously accepted Seagreen 1 EIA (Optimised Project Addendum 2018). All values used followed the Scoping Opinion and the agreement reached at the Ornithology Road Map 6 meeting (10t May 2022).
414. Flight type was set as flapping for all species except gannet, which was set to gliding following advice from NatureScot in their Scoping Consultation response (7 December 2021).
415. Further details on the biological parameters used for CRM are presented in volume 3, appendix 11.3.

Avoidance Rates

416. For the deterministic Band model, avoidance rates for all species were sourced from the SNCBs joint response on approved avoidance rates (SNCBs, 2014; Cook et al., 2014) ( Table 11.45   Open ▸ ). Use of SNCBs (2014) avoidance rates for the primary CRM assessment was advised in the Scoping Opinion (4 February 2022). In addition, an avoidance rate of 0.980 for gannet was also presented for context, following RSPB’s consultation representation, as specified in the Scoping Opinion.
417. There are no SNCBs endorsed avoidance rates for kittiwake or gannet for the extended Band model (Option 3). Therefore, avoidance rates from Bowgen and Cook (2018) were used for comparison, noting that an avoidance rate for use in the extended model is not provided.
418. For the sCRM, avoidance rates for kittiwake, gannet, herring gull and lesser black-backed gull were taken from Bowgen and Cook (2018). SNCBs advice on their preferred avoidance rates for sCRM was not available, but agreement to use rates from Bowgen and Cook (2018) was obtained through the Ornithology Road Map process and confirmed in the Scoping Opinion (4 February 2022). Avoidance rates for sCRM for common and Arctic terns, little gull and great skua were set at 0.980, which followed SNCB advice (SNCBs, 2014).

Table 11.45:  Avoidance rates (± 2 SD) used for Deterministic Basic (Options 1 and 2) and Extended (Option 3) Band Model (2012) (SNCBs, 2014), and sCRM (with 95% Confidence Intervals) (Bowgen and Cook 2018)

1 Values in brackets are ± Standard Deviation.
2 Values in brackets are 95% confidence limits.

419. It should be noted that the avoidance rate of 0.989 recommended for gannet by SNCBs (2014) does not account for macro-avoidance and so there is a case for incorporating an additional macro-avoidance rate for this species, which would reduce collision estimates substantially.
420. Further details on the avoidance rates used for CRM are presented in volume 3, appendix 11.3.

Flight height

421. It was agreed through the Ornithology Road Map process (RM4, 8 December 2021) that the CRM should utilise the generic modelled flight heights from Johnston et al. (2014a; 2014b) for the primary assessment (Band Option 2 and 3). These flight height data were collated from seabird surveys at 32 offshore wind farms in the UK and Europe. Most surveys were boat-based, with height measurements taken visually and assigned to height bands, to derive continuous flight height distributions for 25 seabird species. Further details on the flight heights used for CRM are presented in volume 3, appendix 11.3.
422. In addition, collision estimates for kittiwake based on site-specific boat-based flight heights from observer and rangefinder are presented in volume 3, appendix 11.3 annex B, for context. Compared to estimated annual number of collisions using the generic flight height data for kittiwake for the Developer Approach and the Scoping Approach, the results from using site-specific kittiwake flight heights from rangefinder and visual observer data were considerably lower. This illustrates that the CRM estimates for kittiwake based on the generic flight height data is likely to be precautionary, and this should be kept in mind when reviewing the below results.

Worst-Case Collision Estimates

423. Collision estimates for the worst-case design scenario (307x14 MW wind turbines) for the eight key species are presented in Table 11.46   Open ▸ . Estimated collisions for the Developer Approach (mean densities) and Scoping Approach (maximum densities) are presented. Estimates are rounded to nearest whole bird, apart from for great skua, where very low annual collision numbers were estimated, considerably less than one bird.
424. Relevant avoidance rates used are shown, along with outputs using the sCRM model for comparison. For the sCRM outputs, the mortality estimates for the ‘equivalent’ maximum design scenario are provided, but the scenario is not entirely equivalent to the Band model maximum design due to the different avoidance rates used.
425. For the Developer Approach, results from the sCRM for kittiwake were considerably lower (-46%). Similarly, sCRM estimates were also lower for lesser black-backed gull (-33%) and herring gulls (-58%) unchanged for common tern, and higher for Arctic tern (+43%), little gull (+80%) and great skua (+83%). A similar pattern was also obtained when using the Scoping Approach. The results from the sCRM were lower for kittiwake, herring gull and lesser black-backed gull (-46%, -36%, -33% respectively). For other species, sCRM estimates were unchanged for common tern, and higher for Arctic tern (+36%), little gull (+64%) and great skua (+65%).
426. Due to its stochastic nature, estimates from the sCRM are not directly comparable with Band outputs because the output is a distribution rather than a single estimate of collisions. Recommended avoidance rates also differ between Band and sCRM methods. Further outputs are presented in volume 3, appendix 11.3 annex C.

Table 11.46:  Worst-case estimates for each species identified from the deterministic Band CRM using the generic flight height data (Options 2 and 3) and SNCBs (2014) avoidance rates for the Developer Approach and Scoping Approach. Estimates are rounded to nearest whole bird

1 Values in brackets show Standard Deviation for sCRM.

PVA Approach

427. For gannet and kittiwake, a regional PVA of combined predicted collision and displacement mortality was conducted for breeding colonies within multiple SPAs. For herring gull and lesser black-backed gull, a regional PVA of predicted collision mortality was conducted for breeding colonies within multiple SPAs. The species/ SPA combinations modelled were chosen using a threshold approach advised in the Scoping Opinion (MS-LOT, 2022) and confirmed through the Ornithology Roadmap process (Meeting 6, 10 May 2022). Further details of the SPA combinations and impact scenarios used are presented in volume 3, appendix 11.6.
428. For each of these species, results for the 35-year period are presented and discussed below.
429. It should be noted that for seven of the key seabird species considered here, the regional populations as defined in the breeding and non-breeding seasons in this chapter are different (i.e., they derive from a very different composition of source populations/colonies). The PVAs are relevant to the regional population as defined for the breeding season but not to that defined for the non-breeding season (with the exception of herring gull). The PVAs also account for effects on this regional breeding population during both breeding and non-breeding periods. However, overall, the results of the regional PVAs are considered indicative for assessment purposes.
430. The CRM assessments are presented for each species below.

Gannet

431. For the Developer Approach, annual estimated gannet mortality from collision impacts in the Proposed Development was based on mean densities of flying birds recorded on baseline digital aerial surveys. For the Scoping Approach, this was based on maximum densities of flying birds recorded on baseline digital aerial surveys.
432. A complete range of collision numbers for the Proposed Development, and the different design scenarios for both the Developer Approach and the Scoping Approach are presented in volume 3, appendix 11.3.
433. The estimated number of collisions per bio-season for gannet based on the Developer Approach and the Scoping Approach are presented in Table 11.47   Open ▸ . Figures are presented for the breeding season and the autumn and spring migration periods of the non-breeding season, based on the maximum design scenario (307x14 1MW wind turbines). Highest numbers of collisions were predicted for the breeding season, for both approaches, with lower numbers of collisions predicted for the autumn and spring migration periods of the non-breeding season.

Table 11.47:  Estimated Number of Collisions for Gannet by bio-season in the Proposed Development array area for the Worst-case Scenario (SNCBs avoidance rates, wind turbine 14 MW, Option 2) for the Developer Approach and Scoping Approach. Estimates are rounded to nearest whole bird.

434. In addition, monthly estimated collisions based on an avoidance rate of 0.980 for the breeding season (mid-March to September) are presented in Table 11.48   Open ▸ , for context, as requested in the Scoping Opinion. In both Developer and Scoping Approaches, peak collisions were estimated in the second half of the breeding season, between July and September.

Table 11.48: Estimated Collisions for Gannet in the Proposed Development array area based on Avoidance Rate of 0.980, wind turbine 14 MW, Option 2 and Generic Flight Height, in Breeding Season for the Developer Approach and Scoping Approach

*March collision estimates presented are for the entire month. Gannet breeding season is estimated to start in mid-March (NatureScot, 2020), therefore, only half of the collisions for the month of March were counted in the total breeding season collision estimates.

Magnitude of Impact

435. The overall baseline mortality rates were based on age-specific demographic rates and age class proportions as presented in Table 11.21   Open ▸ . The potential magnitude of impact was estimated by calculating the increase in baseline mortality within each bio-season with respect to the regional populations.

Table 11.49: Estimated Collision Mortality for Gannet in the Proposed Development array area by bio-season in Relation to Baseline Mortality, for the Developer Approach

1 Breeding season assessment is for breeding adults only.

Table 11.50: Estimated Collision Mortality for Gannet in the Proposed Development array area by bio-season in Relation to Baseline Mortality, for the Scoping Approach

1 Breeding season assessment is for breeding adults only.

Breeding Season

436. For the Developer Approach in the breeding season, the total estimated number of gannet collisions was 138 birds ( Table 11.47   Open ▸ ). However, this includes non-breeding adults and immature birds, as well as breeding adults. Based on the proportion of immature gannets recorded on digital aerial baseline surveys in the breeding season, 1% of the population present in the breeding season are immature birds ( Table 11.25   Open ▸ ). This would mean that 137 adult gannets and one immature bird are predicted to collide with wind turbines in the breeding season, based on the worst-case design scenario. However, a proportion of adult birds present at colonies in the breeding season will opt not to breed in a particular breeding season. It has been estimated that 10% of adult gannets may be “sabbatical” birds in any particular breeding season (volume 3, appendix 11.6), and this has been applied for this assessment. On this basis, 14 adult gannets were considered to be not breeding and so 123 adult breeding gannets were taken forward for the breeding season assessment.
437. The total gannet regional baseline breeding population is estimated to be 323,836 individuals ( Table 11.9   Open ▸ ). The adult baseline survival rate is estimated to be 0.954 ( Table 11.21   Open ▸ ), which means that the corresponding rate for adult mortality is 0.046. Applying this mortality rate, the estimated baseline mortality of gannets is 14,896 adult birds per breeding season. The additional predicted mortality of 123 breeding adult gannets would increase the baseline mortality rate by 0.826 ( Table 11.49   Open ▸ ).
438. For the Scoping Approach in the breeding season, the total estimated number of gannet collisions was 170 birds ( Table 11.47   Open ▸ ). However, this includes non-breeding adults and immature birds, as well as breeding adults. Assuming that 1% of the population present in the breeding season are immature birds ( Table 11.25   Open ▸ ), then this would mean that 168 adult gannets and two immature birds are predicted to collide with wind turbines in the breeding season, based on the worst-case design scenario. However, it has been estimated that 10% of adult gannets may be “sabbatical” non-breeding birds in any particular breeding season (volume 3, appendix 11.6), and this has been applied for this assessment. On this basis, 17 adult gannets were considered to be not breeding and so 151 breeding adult gannets were taken forward for the breeding season assessment.
439. Applying the adult baseline mortality rate of 0.046, the estimated baseline mortality of gannets is 14,896 adult birds per breeding season. The additional predicted mortality of 151 breeding adult gannets would increase the baseline mortality rate by 1.01% ( Table 11.50   Open ▸ ).

Non-breeding Season – Autumn Migration Period

440. For the Developer Approach in the autumn migration period, the total estimated number of gannet collisions was 13 birds ( Table 11.49   Open ▸ ), however, this includes adult and immature birds. Based on information presented in Furness (2015), in the non-breeding season 45% of the population present are immature birds and 55% of birds are adults This would mean that seven adult gannets and six immature birds are predicted to collide with wind turbines, based on the worst-case design scenario.
441. Based on Furness (2015), the total gannet BDMPS regional baseline population for the autumn migration period is estimated to be 456,298 individuals ( Table 11.9   Open ▸ ). Using the average baseline mortality rate of 0.151 ( Table 11.21   Open ▸ ), the estimated regional baseline mortality of gannets is 68,901 birds in the autumn migration period. The additional predicted mortality of 13 gannets would increase the baseline mortality rate by 0.019% ( Table 11.49   Open ▸ ).
442. For the Scoping Approach in the autumn migration period, the total estimated number of gannet collisions was 18 birds ( Table 11.50   Open ▸ ), however, this includes adult and immature birds. Based on Furness (2015), in the non-breeding season 45% of the population present are immature birds and 55% of birds are adults. This would mean that ten adult gannets and eight immature birds are predicted to collide with wind turbines, based on the worst-case design scenario. The additional predicted mortality of 18 gannets would increase the baseline mortality rate by 0.026% ( Table 11.50   Open ▸ ).

Non-breeding Season – Spring Migration Period

443. For the Developer Approach in the spring migration period, the total estimated number of gannet collisions was two birds ( Table 11.49   Open ▸ ), however, this includes adult and immature birds. Based on Furness (2015), in the non-breeding season 45% of the population present are immature birds and 55% of birds are adults. This would mean that one adult and one immature gannets are predicted to collide with wind turbines, based on the worst-case design scenario.
444. Based on Furness (2015), the total gannet BDMPS regional baseline population for the spring migration period is estimated to be 248,385 individuals ( Table 11.9   Open ▸ ). Using the average baseline mortality rate of 0.151 ( Table 11.21   Open ▸ ), the estimated baseline mortality of gannets is 37,506 birds in the spring migration period. The additional predicted mortality of two gannets would increase the baseline mortality rate by 0.005% ( Table 11.49   Open ▸ ).
445. For the Scoping Approach in the spring migration period, the total estimated number of gannet collisions was three birds ( Table 11.49   Open ▸ ), however, this includes adult and immature birds. Based on Furness (2015), in the non-breeding season 45% of the population present are immature birds and 55% of birds are adults. This would mean that two adult and one immature gannets are predicted to collide with wind turbines, based on the worst-case design scenario. The additional predicted mortality of three gannets would increase the baseline mortality rate by 0.008% ( Table 11.50   Open ▸ ).

Assessment of Collision Mortality throughout the Year

446. Predicted gannet mortality as a result of collision in the Proposed Development array area for all bio-seasons as calculated above, was summed for the whole year.
447. Using the Developer Approach, the predicted theoretical additional annual mortality due to collision was an estimated 138 gannets. This corresponds to an increase in the baseline mortality rate of 0.85% ( Table 11.49   Open ▸ ).
448. Using the Scoping Approach, the predicted theoretical additional annual mortality due to collision was an estimated 172 gannets. This corresponds to an increase in the baseline mortality rate of 1.04% ( Table 11.50   Open ▸ ).
449. For the Developer Approach, the estimated increase in the annual baseline mortality rate was below 1% and was therefore not considered to be significant in EIA terms.
450. For the Scoping Approach, the estimated increase in the annual baseline mortality rate was just over 1% and therefore were considered to be potentially significant in EIA terms. However, NS advice in the Scoping Opinion was that collision and displacement impacts should be considered as additive within the assessment for gannet, therefore these assessments have been combined.

Collision and Displacement Impacts Combined

451. Following NS advice in the Scoping Opinion results from the collision and displacement assessments were combined, using the annual predicted mortality totals for both the Developer Approach and the Scoping Approach ( Table 11.51   Open ▸ and Table 11.52   Open ▸ ).

Table 11.51: Combined Annual Estimated Numbers of Collisions and Displacement Mortality for Gannet for the Developer Approach

Table 11.52: Combined Annual Estimated Numbers of Collisions and Displacement Mortality for Gannet for the Scoping Approach

452. Using the Developer Approach, the predicted theoretical additional annual mortality due to collision and displacement was a combined total of 182 gannets. This corresponds to an increase in the baseline mortality rate of 1.08% ( Table 11.51   Open ▸ ).
453. Using the Scoping Approach, the predicted theoretical additional annual mortality due to collision and displacement was a combined total of between 216 and 299 gannets. This corresponds to an increase in the baseline mortality rate of between 1.27% and 1.70% ( Table 11.52   Open ▸ ).
454. It should be noted that this approach is considered highly precautionary. As highlighted by NS in the NnG Scoping Opinion (Marine Scotland, 2017a), collision risk and displacement are considered to be mutually exclusive impacts, and therefore combining mortality estimates for displacement and collision should be considered extremely precautionary.
455. These combined collision and displacement mortality estimates suggest a potential significant increase in the baseline mortality rate for gannet for both the Developer Approach and the Scoping Approach, therefore PVA analysis was conducted on the gannet regional SPA population.

Summary of Regional PVA Assessment

456. PVA has been carried out on the regional gannet SPA population considering a wide range of displacement and mortality rates and also a range of collision scenarios. The results of the regional PVAs for predicted displacement and collision impacts for the Project alone during the operation phase for the gannet regional SPA population for the 35 year projection is summarised in Table 11.53   Open ▸ . Further details of the PVA methodology, input parameters and an explanation of how to interpret the PVA results can be found in volume 3, appendix 11.6.

Table 11.53: Summary of PVA Displacement and Collision Outputs for Gannet for the Proposed Development array area plus 2 km buffer after 35 years

1 Starting population taken from volume 3, appendix 11.6.
Developer Approach = 70% displacement and 1% mortality throughout year and mean monthly density for CRM.
Scoping Approach A = 70% displacement; 1% displacement mortality throughout year and maximum monthly density for CRM.
Scoping Approach B = 70% displacement; 3% displacement mortality throughout year and maximum monthly density for CRM.

457. For both the with and without Project scenarios, the gannet regional SPA population is predicted to increase over the 35-year period. For the Developer Approach, the end population size with Project scenario was slightly lower than the without Project scenario. There was no predicted difference in the counterfactual of the population growth rate, and the counterfactual of the population size was also very close to 1.000, while the 50th Centile value was close to 50. These values indicate that the PVA did not predict a significant negative effect from the project alone effects of displacement mortality and collision mortality from the Developer Approach on the gannet regional SPA population after 35 years.
458. For Scoping Approach A, the end population size with Project scenario was lower than the without Project scenario. There was no difference in the counterfactual of the population growth rate, and the counterfactual of the population size was also close to 1.000, while the 50th Centile value was close to 50. These values indicate that the PVA did not predict a significant negative effect from the project alone effects of displacement mortality and collision mortality from Scoping Approach A on the puffin regional SPA population after 35 years.
459. For Scoping Approach B, the end population size with Project scenario was lower than the without Project scenario. There was a very slight predicted difference in the counterfactual of the population growth rate, and the counterfactual of the population size was also close to 1.000, while the 50th Centile value was also close to 50. These values indicate that the PVA did not predict a significant negative effect from the project alone effects of displacement mortality and collision mortality from Scoping Approach B on the gannet regional SPA population after 35 years.
460. Based on the results from the displacement and CRM assessments and the combined regional PVA for the Developer Approach and Scoping Approaches A and B, the magnitude of impact on the regional gannet population is considered to be low.

Sensitivity of the Receptor

461. For gannet, there is evidence that gannets show a high degree of avoidance of offshore wind farms. A detailed study (Krijgsveld et al., 2011) using radar and visual observations to monitor the post-construction effects of the Windpark Egmond aan Zee OWEZ established that 64% of gannets avoided entering the wind farm (macro-avoidance) and a similar result (80% macro avoidance) was also observed during a study at the Thanet wind farm (Skov et al., 2018). Leopold et al. (2013) reported that most gannets avoided Dutch offshore wind farms and did not forage within these. Dierschke et al. (2016) concluded that gannets strongly or nearly completely avoid offshore wind farms.
462. In addition, the Year 1 post-construction study report for Beatrice offshore wind farm reported that gannet showed a marked difference in distribution within the wind farm on post-construction surveys than on pre-construction surveys, with only two birds recorded within the wind farm boundary across all post-construction six surveys undertaken in Year 1. Spatial modelling indicated a significant decrease centred on the wind farm and extending towards the coast with no areas of significant increase. Beyond the region of decrease, the density in the remainder of the survey area was almost identical when comparing pre- and post-construction data (MacArthur Green, 2021).
463. Gannet sensitivity to displacement is discussed in paragraph 209 onwards. Based on evidence from other operational offshore wind farms and a review of gannet GPS tracking data from the Bass Rock, it is considered that the majority of adult gannets passing through the Proposed Development are in transit rather than actively foraging. In addition, the home range of birds breeding on the Bass Rock is very large, in relation to the size of the Proposed Development, while gannets are also known to feed on a wide range of prey species.
464. Based on evidence from post-construction studies, it is considered that collision impacts as estimated for the CRM assessment for gannet are likely to be over-estimates, as it is highly likely that the majority of gannets will avoid the Proposed Development. The first year of post-construction monitoring at Beatrice Offshore Wind Farm recorded virtually no gannets within the wind farm, and concluded that the current collision avoidance rate of 98.9% used in CRM may well be an underestimate of the level of avoidance this species performs (MacArthur Green, 2021).
465. On the basis of these results, which highlight the high degree of avoidance of wind turbines, gannet sensitivity to collision and displacement impacts from operational offshore wind farms is considered to be medium ( Table 11.16   Open ▸ ).
466. In addition, estimated numbers of gannets recorded within the Proposed Development would qualify as nationally important in the breeding season (See volume 3, appendix 11.1, annex G), with individuals likely originating from a number of SPAs in the region. On this basis the conservation importance for gannet was considered to be medium.

Significance of the Effect

467. For combined displacement and collision effects on gannet from the Project alone, for the Developer Approach, the magnitude of the impact is deemed to be low, and the sensitivity of the receptor is considered to be medium. The effect will, therefore, be of minor adverse significance, which is not significant in EIA terms.
468. For Scoping Approach A, the magnitude of the impact is deemed to be low, and the sensitivity of the receptor is considered to be medium. The effect will, therefore, be of minor adverse significance, which is not significant in EIA terms.
469. For Scoping Approach B, the magnitude of the impact is deemed to be low, and the sensitivity of the receptor is considered to be medium. The effect will, therefore, be of minor adverse significance, which is not significant in EIA terms.

Secondary and Tertiary Mitigation and Residual Effect

470. No offshore and intertidal ornithology mitigation is considered necessary because the likely effect in the absence of further mitigation (beyond designed in measures outlined in section 11.10) is not significant in EIA terms. Therefore, the residual impact is considered to be of minor adverse significance, which is not significant in EIA terms.

Herring Gull

471. For the Developer Approach, annual estimated herring gull mortality from collision impacts in the Proposed Development was based on mean densities of flying birds recorded on baseline digital aerial surveys. For the Scoping Approach, this was based on maximum densities of flying birds recorded on baseline digital aerial surveys.
472. For assessment purposes, the breeding season for herring gull has been defined as April to August (NatureScot, 2020). The corresponding non-breeding season for herring gull was based on Furness (2015) but adjusted for overlaps with the previously defined NatureScot breeding season definition, and therefore covered September to March for this species.
473. The estimated number of collisions per bio-season for herring gull based on the Developer Approach and the Scoping Approach are presented in Table 11.54   Open ▸ . Figures are presented for the breeding and non-breeding seasons, based on the worst-case design scenario (307x14 MW wind turbines). For both approaches, highest numbers of collisions were predicted for the breeding season, with lower numbers of collisions predicted for the non-breeding season.
474. A complete range of collision numbers for the Proposed Development, and the different design scenarios for both the Developer Approach and the Scoping Approach are presented in volume 3, appendix 11.3.

Table 11.54:  Estimated Number of Collisions for Herring Gull by Bio-season in the Proposed Development for the Worst-Case Scenario (SNCBs avoidance rates, wind turbine 14 MW, Option 2) for the Developer Approach and Scoping Approach. Estimates are rounded to nearest whole bird.

Magnitude of Impact

475. The overall baseline mortality rates were based on age-specific demographic rates and age class proportions as presented in Table 11.21   Open ▸ . The potential magnitude of impact was estimated by calculating the increase in baseline mortality within each bio-season with respect to the regional populations.

Table 11.55: Estimated Numbers of Collisions for Herring Gull in the Proposed Development array area by Bio-season in Relation to Baseline Mortality, for the Developer Approach

1 Breeding season assessment is for breeding adults only.

Table 11.56: Estimated Numbers of Collisions for Herring Gull in the Proposed Development array area by Bio-season in Relation to Baseline Mortality, for the Scoping Approach

1 Breeding season assessment is for breeding adults only.

Breeding Season

476. For the Developer Approach in the breeding season, the total estimated number of herring gull collisions was 26 birds ( Table 11.54   Open ▸ ). However, this includes non-breeding adults and immature birds, as well as breeding adults. Based on the proportion of immature herring gulls recorded on digital aerial baseline surveys in the breeding season, 8% of the population present in the breeding season are immature birds ( Table 11.57   Open ▸ ).

Table 11.57: Proportions of juvenile, immature and adult Herring Gulls recorded on Digital Aerial Surveys

477. This would mean that 24 adult herring gulls and two immature birds are predicted to collide with wind turbines in the breeding season, based on the worst-case design scenario. However, a proportion of adult birds present at colonies in the breeding season will opt not to breed in a particular breeding season. It has been estimated that 35% of adult herring gulls may be “sabbatical” birds in any particular breeding season (volume 3, appendix 11.6), and this has been applied for this assessment. On this basis, eight adult herring gulls were considered to be not breeding and so 16 breeding adult herring gulls were taken forward for the breeding season assessment.
478. The total herring gull regional baseline breeding population is estimated to be 29,600 individuals ( Table 11.9   Open ▸ ). However, it should be noted that this figure is considered likely to be an under-estimate due to limited surveys of urban gull colonies, which have increased in the region in recent years (Welch, 2019a). A larger regional population would result in a corresponding larger figure for the estimated regional baseline mortality figure, and therefore a lower predicted increase in additional mortality, and this should be borne in mind for this assessment.
479. The adult baseline survival rate is estimated to be 0.878 ( Table 11.21   Open ▸ ), which means that the corresponding rate for adult mortality is 0.122. Applying this mortality rate, the estimated regional baseline mortality of herring gulls is 3,611 adult birds per breeding season. The additional predicted mortality of 16 breeding adult herring gulls would increase the baseline mortality rate by 0.44% ( Table 11.55   Open ▸ ).
480. For the Scoping Approach in the breeding season, the total estimated number of herring gull collisions was 43 birds ( Table 11.54   Open ▸ ). However, this includes non-breeding adults and immature birds, as well as breeding adults. Based on the proportion of immature herring gulls recorded on digital aerial baseline surveys in the breeding season, 8% of the population present in the breeding season are immature birds ( Table 11.57   Open ▸ ). This would mean that 40 adult herring gulls and three immature birds are predicted to collide with wind turbines in the breeding season, based on the worst-case design scenario.
481. As above, a sabbatical rate of 35% for non-breeding adult herring gulls (volume 3, appendix 11.6) has been applied for this assessment. On this basis, 14 adult herring gulls were considered to be not breeding and so 26 breeding adult herring gulls were taken forward for the breeding season assessment.
482. Applying the adult baseline mortality rate of 0.122, the estimated baseline mortality of herring gulls is 3,611 adult birds per breeding season. The additional predicted mortality of 26 breeding adult herring gulls would increase the baseline mortality rate by 0.72% ( Table 11.56   Open ▸ ).

Non-breeding Season

483. For the Developer Approach in the non-breeding season, the total estimated number of herring gull collisions was four birds ( Table 11.55   Open ▸ , however, this includes adult and immature birds. Based on information presented in Furness (2015), in the non-breeding season 52% of the population present are immature birds and 48% of birds are adults. This would mean that two adult and two immature herring gulls are predicted to collide with wind turbines in the non-breeding season, based on the worst-case design scenario.
484. Scoping Opinion advice for herring gulls was to use the regional breeding population within mean maximum foraging range +1S.D (29,600 birds). as the reference population for the non-breeding season. However, a correction factor was required to account for the influx of continental breeding birds into eastern Scotland/UK in the non-breeding season. At the road map meetings, MSS advised (volume 3, appendix 11.8) that this correction factor should be calculated from the proportions of overseas and western UK birds in the UK North Sea and Channel BDMPS (Furness 2015). This correction factor was calculated to be 0.67 (volume 3, appendix 11.5), which results in an additional 19,832 herring gulls as the estimated influx of continental breeding birds. The total herring gull regional baseline population in the non-breeding season, is therefore estimated to be 49,432 individuals. Using the average baseline mortality rate of 0.141 ( Table 11.21   Open ▸ ), the estimated regional baseline mortality of herring gulls is 6,970 birds in the non-breeding season. The additional predicted mortality of four herring gulls would increase the baseline mortality rate by 0.06% ( Table 11.55   Open ▸ ).
485. For the Scoping Approach in the non-breeding season, the total estimated number of herring gull collisions was seven birds ( Table 11.54   Open ▸ ), however, this includes adult and immature birds. Based on Furness (2015), 52% of the population present in the non-breeding season are immature birds, then this would mean that three adult and four immature herring gulls are predicted to collide with wind turbines in the non-breeding season, based on the worst-case design scenario. The regional baseline mortality of herring gulls is estimated to be 6,970 birds in the non-breeding season. The additional predicted mortality of seven herring gulls would increase the baseline mortality rate by 0.10% ( Table 11.56   Open ▸ ).

Assessment of Collision Mortality throughout the Year

486. Predicted herring gull mortality as a result of collision in the Proposed Development array area for all bio-seasons as calculated above, was summed for the whole year.
487. Using the Developer Approach, the predicted theoretical additional annual mortality due to collision was an estimated 20 herring gulls. This corresponds to an increase in the baseline mortality rate of 0.50% ( Table 11.55   Open ▸ ).
488. Using the Scoping Approach, the predicted theoretical additional annual mortality due to collision was an estimated 33 herring gulls. This corresponds to an increase in the baseline mortality rate of 0.82% ( Table 11.56   Open ▸ ).
489. For both the Developer Approach and Scoping Approach, the estimated increases in the annual baseline mortality rate were below 1% and were therefore not considered to be significant in EIA terms.
490. Although these collision mortality estimates did not suggest a potentially significant increase in the baseline mortality rate for herring gull for either the Developer Approach or the Scoping Approach, PVA analysis was conducted on the herring gull regional SPA population.

Summary of Regional PVA Assessment

491. PVA has been carried out on the regional herring gull SPA population considering a range of collision scenarios. The results of the PVA for predicted collision impacts for the Project alone during the operation phase for the herring gull regional SPA population for the 35-year projection is summarised in Table 11.58   Open ▸ . Further details of the PVA methodology, input parameters and an explanation of how to interpret the PVA results can be found in volume 3, appendix 11.6.

Table 11.58: Summary of PVA Collision Outputs for Herring Gull for the Proposed Development array area after 35 years

1 Starting population taken from volume 3, appendix 11.6.
Developer Approach = CRM based on mean monthly density.
Scoping Approach = CRM based on maximum monthly density.

492. For both the with and without Project scenarios, the herring gull regional SPA population is predicted to increase over the 35-year period. For the Developer Approach, the end population size with Project scenario was slightly lower than the without Project scenario. There was no predicted difference in the counterfactual of the population growth rate, and the counterfactual of the population size was also very close to 1.000, while the 50th Centile value was close to 50. These values indicate that the PVA did not predict a significant negative effect from the project alone effects of collision mortality from the Developer Approach on the herring gull regional SPA population after 35 years.
493. For the Scoping Approach, the end population size with Project scenario was lower than the without Project scenario. There was a very slight predicted difference in the counterfactual of the population growth rate, and the counterfactual of the population size was also close to 1.000, while the 50th Centile value was close to 50. These values indicate that the PVA did not predict a significant negative effect from the project alone effects of collision mortality from Scoping Approach A on the herring gull regional SPA population after 35 years.
494. Based on the results from the collision assessment and the regional PVA assessment for both the Developer Approach and the Scoping Approach, the magnitude of collision impacts on the regional SPA herring gull population is negligible.

Sensitivity of the Receptor

495. A review of post-construction studies of seabirds at offshore wind farms in European waters concluded that herring gull was one of the species that showed a weak attraction to offshore wind farms (Dierschke et al., 2016). A review of vulnerability of Scottish seabirds to offshore wind turbines ranked herring gull with the second highest score in the context of collision impacts, based on flight activity at blade height, manoeuvrability, time spent in flight, nocturnal flight activity and conservation importance (Furness and Wade, 2012). Similarly, Furness et al., (2013) scored herring gull as the species of highest concern in the context of collision impacts, while Bradbury et al., (2014), classified the herring gull population vulnerability to collision mortality as very high.
496. On this basis, herring gull sensitivity to collision from operational offshore wind farms is considered to be very high ( Table 11.16   Open ▸ ).
497. In addition, estimated numbers of herring gulls recorded within the Proposed Development would occasionally qualify as nationally important in the breeding season (See volume 3, appendix 11.1, annex G), with individuals likely originating from a number of SPAs and non-SPAs in the region. On this basis the conservation importance for herring gull was considered to be medium.

Significance of the Effect

498. For collision effects on herring gull from the Project alone, for the Developer Approach, the magnitude of the impact is deemed to be negligible, and the sensitivity of the receptor is considered to be very high. The effect will, therefore, be of minor adverse significance, which is not significant in EIA terms.
499. For the Scoping Approach, the magnitude of the impact is deemed to be negligible, and the sensitivity of the receptor is considered to be very high. The effect will, therefore, be of minor adverse significance, which is not significant in EIA terms.

Secondary and Tertiary Mitigation and Residual Effect

500. No offshore and intertidal ornithology mitigation is considered necessary because the likely effect in the absence of further mitigation (beyond designed in measures outlined in section 11.10) is not significant in EIA terms. Therefore, the residual impact is considered to be of minor adverse significance, which is not significant in EIA terms.

Lesser Black-backed Gull

501. For the Developer Approach, annual estimated lesser black-backed gull mortality from collision impacts in the Proposed Development array area was based on mean densities of flying birds recorded on baseline digital aerial surveys. For the Scoping Approach, this was based on maximum densities of flying birds recorded on baseline digital aerial surveys.
502. The estimated number of collisions per bio-season for lesser black-backed gull based on the Developer Approach and the Scoping Approach are presented in Table 11.59   Open ▸ . Figures are presented for the breeding and non-breeding seasons, based on the worst-case design scenario (307x14 MW wind turbines).
503. For assessment purposes, the breeding season for lesser black-backed gull has been defined as mid-March to August (NatureScot, 2020). As no lesser black-backed gull collisions were predicted for the non-breeding season for either the Developer Approach or the Scoping Approach, no further assessment was undertaken for this period.
504. A complete range of collision numbers for the Proposed Development, and the different design scenarios for both the Developer Approach and the Scoping Approach are presented in volume 3, appendix 11.3.

Table 11.59:  Estimated number of collisions for Lesser Black-backed Gull by bio-season in the Proposed Development for the Worst-Case Scenario (SNCBs avoidance rates, wind turbine 14 MW, Option 2) for the Developer Approach and Scoping Approach. Estimates are rounded to nearest whole bird.

Magnitude of Impact

505. The overall baseline mortality rates were based on age-specific demographic rates and age class proportions as presented in Table 11.21   Open ▸ . The potential magnitude of impact was estimated by calculating the increase in baseline mortality within each bio-season with respect to the regional populations.

Table 11.60: Estimated Numbers of Collisions for Lesser Black-backed Gull in the Proposed Development array area by bio-season in Relation to Baseline Mortality for the Developer Approach

1 Breeding season assessment is for breeding adults only.

Table 11.61: Estimated Numbers of Collisions for Lesser Black-backed Gull in the Proposed Development array area by bio-season in Relation to Baseline Mortality for the Scoping Approach

1 Breeding season assessment is for breeding adults only.

Breeding Season

506. For the Developer Approach in the breeding season, the total estimated number of lesser black-backed gull collisions was six birds ( Table 11.59   Open ▸ ). However, this includes non-breeding adults and immature birds, as well as breeding adults. Based on the proportion of immature lesser black-backed gulls recorded on digital aerial baseline surveys in the breeding season, 9% of the population present in the breeding season are immature birds ( Table 11.62   Open ▸ ).

Table 11.62: Proportions of Juvenile, Immature and Adult Lesser Black-backed Gulls Recorded in the Breeding Season on Digital Aerial Surveys

507. This would mean that five adult lesser black-backed gulls and one immature bird are predicted to collide with wind turbines in the breeding season, based on the maximum design scenario. However, a proportion of adult birds present at colonies in the breeding season will opt not to breed in a particular breeding season. It has been estimated that 35% of adult lesser black-backed gulls may be “sabbatical” birds in any particular breeding season (volume 3, appendix 11.6), and this has been applied for this assessment. On this basis, two adult lesser black-backed gulls were considered to be not breeding and so three breeding adult lesser black-backed gulls were taken forward for the breeding season assessment.
508. The total lesser black-backed gull regional baseline breeding population is estimated to be 13,994 individuals ( Table 11.9   Open ▸ ). However, it should be noted that this figure is considered likely to be an under-estimate due to limited surveys of urban gull colonies, which have increased in the region in recent years (Welch, 2019b). A larger regional population would result in a corresponding larger figure for the estimated regional baseline mortality figure, and therefore a lower predicted increase in additional mortality, and this should be borne in mind for this assessment.
509. The adult baseline survival rate is estimated to be 0.913 ( Table 11.21   Open ▸ ), which means that the corresponding rate for adult mortality is 0.087. Applying this mortality rate, the estimated regional baseline mortality of lesser black-backed gulls is 1,217 adult birds per breeding season. The additional predicted mortality of three adult lesser black-backed gulls would increase the baseline mortality rate by 0.25% ( Table 11.60   Open ▸ ).
510. For the Scoping Approach in the breeding season, the total estimated number of lesser black-backed gull collisions was nine birds ( Table 11.59   Open ▸ ). However, this includes non-breeding adults and immature birds, as well as breeding adults. Based on the proportion of immature lesser black-backed gulls recorded on digital aerial baseline surveys in the breeding season,9% of the population present in the breeding season are immature birds ( Table 11.62   Open ▸ ). This would mean that eight adult lesser black-backed gulls and one immature bird are predicted to collide with wind turbines, based on the worst-case design scenario.
511. As above, a sabbatical rate of 35% for non-breeding adult lesser black-backed gulls (volume 3, appendix 11.6) has been applied for this assessment. On this basis, three adult lesser black-backed gulls were considered to be not breeding and so five breeding adult lesser black-backed gulls were taken forward for the breeding season assessment.
512. The regional baseline mortality of lesser black-backed gulls is estimated to be 1,217 adult birds per breeding season. The additional predicted mortality of five adult lesser black-backed gulls would increase the baseline mortality rate by 0.41% ( Table 11.61   Open ▸ ).