1 INTRODUCTION

Contemporary ornithological baseline monitoring (2019-2021) has been undertaken to inform the EIA process for Berwick Bank Wind Farm using digital video aerial surveys completed by HiDef Aerial Surveying Ltd. In order to supplement the baseline data, a suite of boat-based surveys have been carried out with the primary aim of collecting flight height data from the Berwick Bank Wind Farm area to inform and support the EIA.

The previous boat-based surveys carried out in 2009 to 2011 in the Firth of Forth Round 3 Zone (Seagreen 2012a,b) followed a modified European Seabirds at Sea (ESAS) protocol and recorded bird flight heights in three height bands (<20 m, 20 m to 100 m and >120 m) to reflect positions of birds below, at and above the rotor swept height of candidate wind turbines under consideration at that time. Whilst useful to determine the overall proportion of flight activity below 20 m, these bands do not however, directly reflect the potential rotor swept heights of modern or future wind turbine technology. Further boat-based surveys have therefore been undertaken to collect site-specific flight height and behavioural data to inform the Berwick Bank collision risk modelling (CRM).

This report details the methods and results of the 2020 and 2021 boat-based surveys for Berwick Bank Wind Farm.

2 Methods

2.1.1           Survey Design

Boat-based surveys were carried out by ECON Ecological Consultancy Ltd (ECON hereafter) and RPS in the breeding season in 2020 and 2021.  Surveys were planned to start in June 2020; however, due to the COVID-19 outbreak, they were started in July 2020 and continued monthly in August 2020 and in April, May and June 2021.

The survey in 2020-2021 followed a modified methodology based on COWRIE recommendations (Camphuysen et al., 2004) for ornithological surveys of wind farms that stems from the ESAS protocol (see Tasker et al., 1984, Webb & Durinck, 1992), as utilised for the previous boat-based surveys in the Firth of Forth Round 3 Zone (Seagreen, 2012a,b, 2018). Surveys were undertaken from the Artemis (IMO reference 9119713), a 28 m Scottish Fisheries Federation vessel which provides a suitable viewing platform and meets the vessel requirements as set out by COWRIE (Camphuysen et al., 2004).

Methodological modifications involved repeatedly sampling locations along a continuous transect route throughout the day, broadly following the method documented in Embling et al. (2012). The survey design for Berwick Bank consisted of four survey sites ( Figure 2.1   Open ▸ ), with each site having two parallel 15 km line transects separated by a 3 km gap. The sites were selected based on the analysis of the baseline boat-based survey data from Seagreen Phase 1 baseline surveys (2009-2011) (and prior to the redesign of the Berwick Bank project as detailed in the October 2021 Scoping Report) to achieve good representative spatial coverage across the site. Each 15 km transect was surveyed at approximately 8 knots, which took around one hour to complete and it took approximately 2.5 hours to complete the survey site and return to the start location. The cross lines between the two transects were not surveyed to ensure the lines remained independent, that there were no variations in survey effort or overlap and that there was the consistent orientation of survey lines.

Figure 2.1 Boat-based transect routes to survey the four sites. At the time of survey, sites were known as Berwick Bank (sites 1 and 2) and Marr Bank (sites 3 and 4).

Each survey site ( Figure 2.1   Open ▸ ) was repeatedly surveyed over the course of one day, aiming for four repetitions at dawn, morning, afternoon and ending in the evening, with each site surveyed once a month in July and August 2020 and April, May and June 2021, in order to cover the core breeding period of the sensitive seabird receptors. The amount of time between repeat transect runs varied due to the amount of daylight hours, with the time between runs used to collect additional flight height data using the rangefinder and for convenience breaks for the surveyors. On some days, adverse weather constraints prevented full coverage of four survey repetitions.

Transects were consistently surveyed in a clockwise direction to ensure continuity in snapshots between surveys. However, the start location was varied between surveys to try to prevent potential bias associated with the position along the transect. The order in which these four survey sites were visited was randomly selected to again ensure consistency with the previous surveys of the Firth of Forth Round 3 Zone in 2009-2011 and surveys of Seagreen Phase 1 in 2017. The exact timing of surveys within each month were determined by the first identified ‘window’ of suitable weather forecasted.

The use of two surveyors, one either side of the vessel, and a dedicated data recorder for all surveys enabled the repeat sampling of sites throughout the day between sunrise and sunset and allowed investigation of any changes in bird densities to occur. All data collected were then considered for use in Distance analysis of observations of birds on the water (Buckland et al., 2001).

2.2 Data Collection

Birds in flight were quantified through radial snapshots at pre-determined locations, and birds on the water used the line transect method. Birds in flight were also recorded continuously throughout the survey. All birds were assigned to distance bands perpendicular from the boat noting the side of the vessel (port or starboard); A: 0-50 m, B: 50-100 m, C: 100-200 m, D: 200-300 m, E: >300 m. Distance bands A to D were classed as within transect, band E was outside the transect. Sea state, wind speed and direction and subjective surveyor defined visibility scores were also recorded throughout the surveys.

Age and plumage data were also collected as part of the standard survey design to enable partitioning of birds to (presumed) breeding and non-breeding groups. Birds were aged by calendar year from 1 to 6 and all observed plumage characteristics were noted.

As it is not deemed possible to consistently separate non-breeding adult auks from advanced juvenile birds from at-sea observations, auk species in the breeding season that are not in summer (breeding) plumage and not obviously non-adult, were noted as being in winter plumage.

In addition to age and plumage, the following information was recorded and described for individual or grouped records whenever possible:

• All interactions with other birds (of defined species) and/or marine mammals;
• Any perceptible behaviour (e.g., sleeping, preening);
• Interactions with the survey vessel or any other vessel (e.g., escape diving or circling);
• Foraging or feeding activity and interactions, including within defined and described mixed species feeding aggregations (MSFA);
• Presence of post-feeding aggregations;
• Birds carrying fish (with fish being identified to the lowest taxonomic rank possible); and
• Associations with features of interest such as tidal fronts, upwellings, currents, detritus, flotsam and jetsam etc.

Every effort was made to record as much behavioural detail as possible, especially in relation to foraging but also other behaviours such as chick guarding etc (see Camphuysen et al., 2004). However, it must be noted that at times, high densities of birds can make fine scale recording of behaviours at the individual level impossible. In these cases, broad categorisations are applied, often to grouped birds of the same species with shared characteristics (e.g., distance band, flight height and direction etc).

To shed light on the potential environmental factors that can contribute to ecosystem functioning within the Berwick Bank wind farm area, a low-cost monitoring solution to provide continuous measurements of surface waters was utilised. A Star Oddi conductivity temperature depth (CTD) instrument (3-37 mS/cm, 1-1000m) was deployed and towed at the surface to monitor water temperature, salinity, and conductivity at a measured depth (results are provided within Appendix 4).

2.2.1           Flight Height

Flight height data collection followed the methods applied during the surveys of the Seagreen Phase 1 Wind Farm in 2017 (Seagreen, 2018) and published by Harwood et al. (2018). All surveyors visually estimated flight heights of all birds seen in snapshots or line transect in 5 m bands (i.e., >0-5, >5-10, >10-15, etc). Surveyors also utilised optical laser rangefinders (Nikon Forestry Pro II) to provide flight height measurements when they were not surveying the line transect or if very few birds were present. Both data sets were analysed to derive flight height distributions for comparison with each other and with the generic flight height distribution published by Johnston et al. (2014). All surveyors received training in the correct use of the rangefinders prior to the start of the surveys.

The survey team included a dedicated surveyor tasked with collecting as many rangefinder flight height observations as possible. To increase the capacity for rangefinder flight height observations, two additional laser rangefinders were used opportunistically by the other observers in the survey team. However, surveyors always prioritised the basic line transect survey to ensure that that bird abundances were not underestimated. Where possible, a GPS waypoint was associated with records to allow spatial referencing.

The dedicated rangefinder operator continually scanned for potential seabird targets. As soon as a potential target bird appeared, attempts were made to measure its flight height. The rangefinder operator aimed to take a measurement of every bird that came into range of the instrument (the Nikon Forestry Pro II has a published measurement range of 7.5 – 1,600 m); most birds were targeted up to approximately 400 distance, with infrequent measurements of birds up to 1 km away and two exceptional records up to approximately 2 km away. Due to discrepancies in size, the range at which larger species such as gannet or great black-backed gull can be measured is greater than that for smaller species such as terns (larger species could be detected up to 1 km horizontal distance, whereas smaller species were recorded within 400 m from the boat). When a range of species were present, preference was given to kittiwake and gannet, the key species to be assessed using CRM. However, due to those species occurring so frequently and being well represented in the data, novel species (e.g. terns and skuas) were also selected if the opportunity arose. Typically, there were relatively few potential targets available at any one time, so data were collected for most birds that came within range. However, when swamping occurs the primary objective was to acquire as many verified flight heights as possible and targets deemed most likely to yield data were selected (i.e. that can be easily targeted with the rangefinder).

2.3 Data Analysis

2.3.1           Abundance

Seabird density estimates were derived using Distance analysis on the visual observational data to account for birds undetected during the boat-based surveys (due to detection decreasing with increasing distance from the transect line; Buckland et al., 2001). Distance analysis was carried out on the observed counts for the five species: kittiwake, gannet, guillemot, razorbill and puffin. This analysis was carried out in R Studio (R Core Team, 2021), using the Distance package (Miller, 2019) and involved fitting either half-normal or hazard-rate detection functions, the model chosen based on Akaike’s Information Criterion (AIC) score (Miller et al., 2019). 

Additionally, as auk species dive underwater to forage, birds are likely to be missed during the surveys and so a correction factor was applied to sitting guillemot, razorbill and puffin to account for this. The correction factor for each species were taken from Thaxter et al. (2010), with the correction factor derived from time spent under water during the chick-rearing stage. The correction factor used for guillemot sitting on the sea surface was 1.375, razorbill 1.174, and puffin 1.1416.

To generate densities across the survey sites, once distance corrected and adjusted for missed auk species, abundances were divided by the total number of cells and snapshots surveyed by each site: snapshots were conducted at 500 m intervals (~ 2-minute intervals) rather than timed along the specified transect lines, thereby compensating for any changes in vessel speed according to variable currents or sea state within and between surveys, with all transects designed to be divisible into complete 500 m cells (sites therefore contained 60 cells).

2.3.2           Flight Height 

Following Cook et al. (2018), Normal, log-normal, gamma, normal-mixture and gamma-mixture distribution were fitted to the rangefinder data using the mixtools (Benaglia et al., 2009) and fitdistrplus (Delignette-Muller and Dutang, 2015) packages in the R statistical package (R Core Team, 2021). The best fitting distribution was selected based on visual appraisal (Appendix 1).

From the visual observation height estimates, the proportion of records in each 5 m height band were categorised, which will allow an overall estimate of the proportion of birds deemed to be below and at rotor-swept height for use in CRM.

All flight heights were corrected to account for variation in observer eye height. The negative values arising from this analysis were due to variations in swell height influencing the eye height of the observer and for the purposes of this initial analysis were corrected to 1 m above sea level as they will be low-flying birds.

To investigate if distance from the boat had an influence on observed flight heights, flight heights from both rangefinder and surveyor flight height estimates were compared against distance of the observation from the boat. This was done to examine if the presence of the vessel potentially had an influence on the flight heights being recorded (e.g., if birds fly lower when close to the boat, flight heights could be underestimated); species such as kittiwake may be attracted to the boat (Wahl and Heinemann, 1979). Seabirds also often display Area Restricted Search (ARS) behaviour (whereby they increase sinuosity and reduce speed in specific areas; Hill et al., 2000) near fishing boats, often foraging behind them (Votier et al., 2004). By comparing flight height against distance, it can provide insight into whether those flight heights recorded are associated with such behaviour and hence maybe influence the distribution of flight heights being recorded.

3 Results

3.1.1           Surveys

Table 3.1 to Table 3.5 provide details of the dates, timing and conditions of each survey in July and August 2020 and April, May and June 2021.

Table 3.1 Daily survey effort and summary of weather conditions for July 2020 boat-based surveys

Table 3.2 Daily survey effort and summary of weather conditions for August 2020 boat-based surveys

Table 3.3 Daily survey effort and summary of weather conditions for April 2021 boat-based surveys

Table 3.4 Daily survey effort and summary of weather conditions for May 2021 boat-based surveys

Table 3.5 Daily survey effort and summary of weather conditions for June 2021 boat-based surveys