VULNERABILITY OF SCOTTISH SEABIRDS TO OFFSHORE WIND

4. DISCUSSION

The key results are ranked species lists in Table 11 (ranked species concern in relation to collision mortality impacts on populations) and in Table 12 (ranked species concern in relation to displacement impacts on populations).

A draft version of the factors and species-scores was sent to seabird experts for comment. Most reviewers suggested no change to the factors used, and no change to most scores. Most reviewers felt that one or two out of the 228 scores should be adjusted, so agreed with more than 99% of the scores. Two scores were identified that were consistently questioned by reviewers and the scores for these species were altered to bring them in line with this consensus opinion. So the scores presented in the current document have broad agreement from a diverse group of relevant European seabird experts. However, there is uncertainty in the scoring, and in the relevance of particular factors. In particular, there was a broad consensus among reviewers that flight height was considerably more important than any of the other factors in assessment of collision risk, and that this factor should also be weighted higher than the conservation importance score. After discussions, and comments from reviewers, we decided to move away from the formula used by Garthe and H◘ppop (2004) to recognise firstly that there is broad support for the view that collision concern should be considered separately from displacement concern, as the rankings of species on these two features are very different. Secondly, flight height is generally considered to be the key factor in the assessment of collision concern, and there are some mixed views among experts over the relative importance of manoeuvrability, percent of time flying, and amount of nocturnal flight in affecting collision risk. Our use of two separate indices and the downweighting of the last three factors recognises this. This approach also seems more appropriate now than it was when Garthe and H◘ppop (2004) prepared their paper, because there are now considerably more detailed data on seabird flight height from the work of Cook et al. (2011). Nevertheless, it is clear that there is an urgent need for more accurate flight height data and a better understanding of how flight height varies according to environmental conditions.

Desholm (2009) not only suggested that empirical evidence did not strongly support the idea that seabirds that fly at night are at higher risk of collision with offshore wind turbines, but also suggested that a suitable index ranking species by level of concern could be constructed from just two factors; conservation concern expressed as the proportion of the biogeographic population passing through the area of risk, and the importance of adult survival rate in determining population trend (technically expressed as 'elasticity' but essentially recognising that additional mortality of long-lived birds will have a greater impact on their population than the same level of additional mortality affecting birds with a naturally low adult survival rate since that tends to be balanced by high reproductive output in species with low adult survival). Desholm (2009) showed that this very simple and statistically robust model worked well for a range of migrant bird species from coal tits to eider ducks passing an offshore wind farm in the Baltic Sea. However, that range of species extends from very short-lived birds (some small passerines are fortunate to live to be one year old, but may rear ten or more chicks) to very long-lived birds such as most seabirds (which typically live tens of years and in many cases do not even start to breed until several years old and rarely rear more than one chick per pair). In our analysis we are dealing mainly with species at the long-lived end of this spectrum of life histories and so the simple model focusing on adult survival rate becomes less useful as there is relatively little variation in this among most seabird species. In addition, it seems lower risk of collision than seabirds that fly at higher altitudes. So the simple model proposed by Desholm (2009) while suitable for broadly ranking birds of all kinds and concluding that seabird populations are of high concern whereas small passerine populations are not, does not perform well when comparing between seabird species of similar demography but differing in ecology and behaviour.

A factor that has been raised as possibly affecting collision risk for seabirds is weather conditions such as fog or heavy rain, which may obscure turbines. Such effects might over-ride any species-specific differences in vulnerability. In reviewing the results of studies at demonstration offshore wind farms in Denmark, Fox et al. (2006) stated ' Waterbird migration typically reduces substantially or ceases during periods of poor visibility and indeed during the observations reported here, the arrival of fog and active rain associated with frontal systems invariably resulted in the cessation of active migration that had been observed during previous periods of good visibility. We must stress that these responses are those by waterbirds generally and at Nysted by common eiders in particular' and ' There has been a general concern world-wide that even if there are few collisions under normal conditions, bird populations may be affected by catastrophic mortality events on rare occasions when visibility is impaired by fog or other adverse weather conditions. The observations at Nysted that waterbirds tend not to fly in the area off the turbines at night, or under adverse weather conditions (as found elsewhere; Petterson 2005) suggest that collision risk is not likely to be high even under conditions when the turbines are less visible'. These observations suggest that catastrophic mortality incidents caused by adverse weather conditions are less likely at offshore wind farms than has been suggested by some. Similarly, seabirds that fly at night might be more at risk of collision on darker nights or during adverse weather. However, even the most highly adapted seabirds to nocturnal flight activity (for example white-chinned petrels Procellaria aequinoctialis) show greatly reduced flight activity at night when there is no moonlight available to guide them (Mackley et al. 2011).

With considerable research effort being put into studying the behaviour of seabirds at sea, and new developments such as use of data loggers to measure flight heights of individual seabirds throughout the breeding season or overwinter, it will be possible to revise the scoring in the light of more detailed data. In particular, there is a need for more accurate data on seabird flight heights while at sea. The influences of local conditions and environmental variation on this parameter also need to be better understood. For example, seabirds searching for fishing vessels may be able to do so more successfully if they fly high so can see greater distances (Skov and Durinck 2001, Furness et al. 2007), whereas seabirds that are commuting to a specific foraging site (such as a sand bank with sandeels, or a predictable frontal region with aggregations of zooplankton) should fly low over the sea surface to minimise travel costs and time (Pennycuick 1987), especially if flying into a headwind (Dierschke and Daniels 2003). So birds of the same species may behave differently if utilizing different feeding opportunities. Similarly, although divers tend to fly relatively low over the sea when moving between feeding sites at sea, divers flying from a nest site to the sea come off the land at a greater height, and often come off land over cliff coastline (M. Heubeck in litt.). In such cases their flight height in the initial part of the foraging trip is dictated by land topography, and so varies by location. In this context, we note that most of the data used by Cook et al. (2011) come from offshore wind farm sites in the southern UK or from overseas, and very little from Scottish waters.

There were some discrepancies between the published data on flight heights of seabirds in Garthe and H◘ppop (2004) and in Cook et al. (2011), and these often differed from data for the same species of seabirds at a 'coastal' offshore wind farm near Blyth (Rothery et al. 2009). Where these discrepancies occurred we made a judgement as to the appropriate score for the species with regard to scores allocated to similar species as well as to the apparent 'outlier' nature of particular data.

This clearly points to a need to obtain larger and more accurate data sets on seabird flight heights, and to be cautious about the ranking of seabird species presented in Table 11. However, we suggest that species with high scores in Table 11 should be given particular concern in relation to offshore wind developments. This table identifies gulls, white-tailed eagles, gannets, skuas and divers as being the groups whose populations are most at risk in a Scottish context. Many seabird species rarely fly at turbine blade height, and so appear to have negligible risk of population level impacts from collision mortality, though it would be desirable to have more data on flight heights to allow this inference to be converted into a confident conclusion that might permit species to be scoped out of assessments. These include sea ducks, alcids, storm-petrels and shearwaters ( Table 11). The low risk for these species is consistent with data from long-established offshore wind farms (ICES 2011).

According to ICES (2011) ' the picture that emerges from functioning marine windfarms is of little observed bird mortality, and a tendency for seabirds to avoid the arrays of turbines when flying past'.

This is consistent with data from Swedish and Danish offshore wind farms. Petterson (2005) recorded only one collision of a sea duck from about 2 million migrating past a Swedish offshore wind farm, while Fox et al. (2006) used radar studies at Nysted to predict a collision rate of 0.02% (i.e. a 99.98% avoidance rate) for 235,000 common eiders migrating past that site, and observed no collisions by infra-red monitoring (of a single turbine). Fox et al. (2006) did observe significant displacement, of scoters in the short term, and of divers without any evidence of habituation. Lindeboom et al. (2011), studying ecological changes at an offshore wind farm in the Netherlands, reported ' gannets, scoters, auks and divers showed strong avoidance behaviour in their flight patternin the vicinity of the farm' and ' gulls, cormorants and terns did not avoid the farm and used it for foraging'. There is clearly a need for a better understanding of the extent to which displacement of seabirds from wind farms does occur, and what population-level effects, if any, arise from this. Meanwhile, in the absence of such research to date, indirect assessments are all that is available to regulators, developers and consultants. In assessing potential importance of displacement for different seabird species ( Table 12), although there was strong consensus among reviewers for the scores used, this consensus may be more a result of uncertainty than confident agreement, and so the ranking of species needs to be treated with caution. However, we suggest that species with scores over 15 (divers, scoters, goldeneye, scaup, eider, black guillemot, Slavonian grebe) should be considered as focal species for concern about potential displacement effects, while species with scores below 8 (fulmar, storm-petrels, shearwaters, gulls, skuas, gannet, little auk, and white-tailed eagle) seem very unlikely to be affected by displacement.

In scoping potential areas for offshore wind farm development in Scottish waters, Davies and Watret (2011) considered constraints implied by seabird SPAs, and the distribution at sea of seabirds as indicated by the European Seabirds at Sea database. These data were combined with the flight height data presented by Cook et al. (2011) to assess numbers of seabirds flying at collision height risk in different parts of the Scottish marine area. The development of sensitivity scoring and conservation importance scoring for individual species of seabirds may help to refine such assessment of environmental constraints by allowing a focus on the seabird species of greatest concern. This would most usefully be combined with mapping of the distribution of seabird SPAs and the numbers of each species protected at these sites. Thus, in addition to improving knowledge of flight heights and the implications of these for collision risk, a useful improvement to the conservation importance score may be possible if an up-to-date database of numbers of seabirds in SPAs in Scotland could be established, and kept up to date, allowing better understanding of the consenting risk for specific developments.