Abstract
This report presents the final results of a four-year study of seabird distribution patterns in and around the first offshore wind farm in Dutch North Sea waters. This wind farm, known as OWEZ (Offshore Wind farm Egmond aan Zee) is situated 10 - 18 km off the Dutch mainland coast, northwest of the port of IJmuiden. Seabirds were repeatedly surveyed along pre-set survey lines, covering a rather large area around OWEZ. This survey design was chosen to make comparisons between the presence of birds within the wind farm and in the surrounding area, while taking into account the general distribution patterns in the general area. The latter were modelled as a function of distance to the coast and north to south clines in density patterns. Temporal differences could be explored by comparing the distribution patterns in one year before construction of the wind farm (T-0 surveys) with three years of post-construction surveys. Both the spatial and the temporal patterns were under influence of other factors than of the wind farm OWEZ alone, however. The Dutch government allowed for a second wind farm to be built in close proximity to OWEZ, in the early years of this study. This second offshore wind farm (known as Princess Amalia Windpark, PAWP) came into operation shortly after OWEZ and the area taken up by this second wind farm should be seen as a second impact area within the larger study area. A third anomaly in the study area is an intensively used anchorage area, where ships destined for IJmuiden port wait to enter. Some 20 ships were usually anchored here; numbers seemed somewhat higher (but went unrecorded) in the last year of the study. Within the general study area, seabirds thus had a choice to go into OWEZ, PAWP, Anchorage or to stay out of these areas, in the remaining, open sea. This reference area surrounding the impact areas was not free from human impacts either, however. Shipping is intensive in Dutch nearshore waters. In the study area shipping comprised traffic approaching or leaving IJmuiden port, ships in transit and fishing. The latter in particular has an influence of the distribution patterns of some seabirds: those scavenging for fishery waste, like gulls. As fishing is not allowed inside the wind farms, the largest concentrations of gulls and allies during the T-1 surveys were likely found in the reference area, where fishing continued. Another large source of variation was changing habitat from closely inshore to further offshore. On top of these rather predictable sources of variation, there was considerable between-years variation for most seabirds and survey months. Such variation usually impacted the whole study area (and probably much larger parts of the North Sea), making year-to-year comparisons, or T-0/T-1 comparisons, more difficult. These sources of variation hindered to some extent the primary aim of this study, which was to determine whether seabirds would be avoiding the wind farm, or be attracted to it, or be indifferent.
Initially, a BACI (Before-After Control Impact) approach was used to address this question, at the level of individual seabird species. First, a series of Before surveys was carried out, covering one whole year (these surveys ran from 2002-2004). Next, three sets of After surveys were carried out (2007-2010). This set-up allowed for comparisons between the Before (construction) and After (the wind farm had become operational) situations, but also between “wind farm” (as well as the other anomalies) and “not-wind farm” within any given survey. The considerable year-to year variation in seabird presence made comparisons between the single set of T-0 surveys and the three sets of subsequent T-1 surveys difficult. Within-survey comparisons were therefore more informative. Within survey comparisons do not rely on T-0/T-1 comparisons, which might be impacted by other factors than the presence of the wind farm, Spatial variation, i.e. general changes in seabird density related to distance to coast and/or northing, were taken into account in these comparisons.
When sufficient data were collected for a given seabird species and month, Generalised Additive Mixed Models (GAMM) or Generalised Additive Models (GAM) were used to explore the relative contributions of location, expressed as distance to shore and latitude and the presence of one of three anomalies or impact areas within the study area: the OWEZ wind farm, the adjacent Princess Amalia Wind Farm, or the Anchorage off IJmuiden, to the distribution patterns found. Presence/absence data, rather than densities of seabirds were used because these data were less affected by both large numbers of zero-counts within the data set and a few counts with very large numbers, or between-observer differences. Between-observer differences were minimised, by using the same principal observers over long time spans and by always using observers in teams of two.
Different results were found for different seabird species. Little impact of the wind farm on most of the so-called nearshore species was found, as these birds rarely ventured out so far to sea, that they would reach OWEZ latitudes. This result is different from Indifference, as the birds concerned simply did not venture out to sea far enough to meet up with the wind farm; this resulted in “Insufficient data” when comparisons between the wind farm and surrounding areas had to be made. This group comprises the Red- and Black-throated Divers, Great Crested Grebe, Common Scoter, Black-headed Gull and “Commic” Terns (Common and Arctic Terns taken together as these could not always be specifically identified). Densities of all these birds at wind farm latitudes were mostly so low, that few individuals were available to fly or swim into the wind farm.
A similar, but mirrored pattern was found in species that mostly occur further offshore, to the west of OWEZ. Densities of Northern Fulmars were always low around OWEZ, most of these birds occurred further west. None were ever seen to enter the wind farm, but ecological consequences of the loss of a small surface area of sea at the fringe of its huge range, must be negligible. Two other birds that tended to occur mostly offshore showed different reactions to the wind farm. Northern Gannets tended to fly around the wind farm, while Black-legged Kittiwakes seemed mostly indifferent to the wind farm.
Large gulls, the most numerous seabirds in the general area, were mostly found associated with fishing vessels. As fishing is no longer allowed in the wind farms, gull numbers were never very high here during the T-1 surveys. Gull distributions were always very patchy around it, as most gulls go where the fishers go. Most gulls seemedrather unconcerned about the presence of offshore turbines, flying through the wind farm without visible behavioural adjustment and resting on the foundation poles of the turbines in small numbers. The main effect of the wind farms on gull distribution patterns is that trawlers are kept at bay and that the largest concentrations of gulls now occur outside the wind farms, around the trawlers that keep working the general area.
Sandwich Terns and Little Gulls occurred throughout the study area while migrating across the study area, and were expected to be able to profit from the presence of the wind farm, by exploiting it for feeding, resting or courtship. These birds reportedly fed in the tidal wakes behind the monopoles of the Danish Horns Rev wind farm (Elsam Engineering & Energi 2005; Elsam Engineering 2005; Petersen & Fox 2007) and are known to extensively use navigational buoys for resting and courtship display in Dutch waters (Tulp & Schekkerman 1997) and were thus expected to also use OWEZ in these respects. However, although both Sandwich Terns (very rarely) and Little Gulls (rarely) were seen inside the wind farm on occasions, most of these birds seemed to prefer flying around the wind farm rather than entering it.
One species, the Great Cormorant, was clearly attracted to the wind farm. Birds from two mainland (coastal) colonies, Zwanenwater (Petten, at 30.3 km from the metmast) and Hoefijzermeer (Castricum, at 18.7 km) were quick to discover that the wind farm provided good offshore feeding and resting conditions. Resting (out of the water) is critically important for cormorants, that need to dry their feathers after feeding bouts under water. Birds commuted between the mainland and OWEZ (and later further on, to PAWP as well) in rather large numbers, while OWEZ and certainly PAWP latitudes were off limits to these birds when no seating was provided.
Auks, in these parts Guillemots and Razorbills, offered the best possibilities to study avoidance from wind farms. Earlier studies, in and around the Horns Rev wind farm, had indicated strong avoidance in auks (Elsam Engineering & Energi 2005; Elsam Engineering 2005; Petersen & Fox 2007). Results for OWEZ were less clear-cut. Both species showed Indifference/Insufficient data in many situations, and Avoidance in some. However, when avoidance was found, this was not total, and Guillemots and Razorbills were both seen inside the wind farm, and also inside the neighbouring wind farm PAWP, with a much higher turbine density. Turbine density probably did have an effect on avoidance though, avoidance being apparently stronger in PAWP (but not 100% either). Measuring the effect of relatively small wind farms on birds that occur in rather low general densities, requires more effort inside the wind farms than was realised in most of our T-1 surveys, due to a rather broad line spacing. Therefore, after an evaluation of the results obtained until 2008 (as outlined in report OWEZ_R_221_T1_20100329_local_birds) more transect lines were introduced in the last set of surveys, and an extra winter survey was carried out in the last year, when auks were present. This approach yielded better results than earlier surveys, but with only one winter’s worth of such data, we still have few statistically significant cases of avoidance. Future work on these species, focussing on the wind farms themselves, is likely to shed more light on the exact amount of disturbance, as a function of both bird density and turbine density.
The data for all species may be summarised as follows (Table 0), given the four possible outcomes of inside/outside wind farm comparisons (with inside wind farm meaning: within the OWEZ perimeter and outside wind farm meaning: outside either OWEZ, PAWP or Anchorage). A total of 17 T-1 surveys were conducted and analysed (see Table 2). Statistical analysis was only possible in situations (bird/month combinations) with sufficient number of birds found within the whole study area and also at longitudes of the wind farm. In other cases, a statistical test of the survey results was not possible: the data were Not applicable. When sufficient birds were available for analysis the outcome of the statistical test was either: Attraction, Avoidance or Non-significance. In Table 0 the numbers of times either result was achieved are summed for all species considered. From this overview it is clear that an effect of the wind farm, in terms of statistically significant Avoidance Attraction could not be demonstrated for most situations, either because the results were Not significant, or because the data were Not applicable. Note, however, that the Not significant category may contain Type II errors due to insufficient statistical power. Attraction was clear in one species, the Great Cormorant. Attraction was also found in some months for several gull species, but gulls also showed Avoidance or Indifference (Non significance) in other situations. Significant Avoidance was found in divers, grebes, gannets, Little Gulls and both auks (Guillemot and Razorbill), but for all of these in only in a minority of surveys with sufficient numbers of birds present.
Average seabird densities (not corrected for birds missed by the observers) during each survey are presented in Appendix 1, separately for the strata OWEZ, PAWP, Anchorage, and the remaining Reference area. Total numbers of all birds and marine mammals observed per survey are given in Appendix 2.