VULNERABILITY OF SCOTTISH SEABIRDS TO OFFSHORE WIND

3.2 Flight altitude (Factor 2)

This factor indicates risk of collision because seabirds that only fly very low over the water will be below the height swept by turbine blades, whereas seabirds that habitually fly high above the water are likely to be at heights that would put them at risk of collision with blades. This factor is widely considered to be of overwhelming importance in determining the risk of collision of seabirds with offshore wind turbines (Band 2011, Cook et al. 2011). While many migrating passerines fly at, or above, turbine height (H◘ppop et al. 2006), Dierschke and Daniels (2003) found that over 90% of all divers, sea ducks, gulls and terns flew at below 50m above sea level, and that these birds tended to fly at lower heights under windy conditions, presumably to reduce costs of flight into headwinds (Dierschke and Daniels 2003). Flight altitude scores and data were taken from two primary sources which carried out detailed reviews: Garthe and H◘ppop (2004), and Cook et al. (2011). These studies can be considered independent, as Cook et al. (2011) neither incorporated data from Garthe and H◘ppop (2004) into their review, nor compared their data with the previous work. Both studies reported flight heights only for birds in flight (i.e. birds sitting on the water were not scored as zero height). There are advantages to using data such as presented by Cook et al. (2011) as these data come from offshore wind farm sites so should be appropriate for seabirds at such locations (though hardly any of their data are from Scottish marine areas). Ideally, this factor would best be presented as reported values of percentages of birds flying at blade height, without collapsing such data into scores on a 5 point scale and so, unlike Garthe and H◘ppop (2004), we have followed that aim by using percentage values for each species. However, although Cook et al. (2011) provide the best available data on this topic, there are a number of problems with the Cook et al. (2011) data, as those authors themselves recognise. Most data are from observations made from boats, and probably include some, and possibly significant numbers of seabirds scared into flight in front of the boat (Camphuysen et al. 2004). Such birds tend to fly low so will bias the distribution of flight heights. Estimates of the height of flying seabirds used by Cook et al. (2011) are mostly very crude, mostly being based on estimates by observers and not on measurements. Data were also recorded not as exact heights, but as numbers of seabirds in different flight height bands, so Cook et al. (2011) fitted distributions to these data to provide model estimates, rather than reporting proportions at blade height from actual data. Some of the model fits were good, but others were not. For some species, sample sizes are very small (for example the 2% of European storm-petrels reported as flying at blade height represent one individual, given that there were data for only 52 European storm-petrels). Some reported differences between species appear to contradict what is known about those particular species. For example, the proportion flying at blade height was three times higher for black-throated divers than for red-throated divers, a difference that seems unlikely to those who know the winter behaviour of these species well. Also, model fit for black-throated diver was not very good, explaining less than half of the variation in the data set. In the case of eiders, Cook et al. (2011) chose to exclude data from three offshore wind farms where eiders were recorded flying high, in order to obtain a better model fit. Exclusion of high flying birds clearly biases the estimate of the proportion of eiders that fly at blade height, so the model estimates for this species appear to be inappropriate for assessing collision risk, at least at sites where eiders may migrate through the area. Cook et al. (2011) suggest that leaving out data for high flying eiders can be justified because eiders at those sites may be migrating, and so flying at a different height which confounded modelling. However, such an argument seems inappropriate, and also overlooks the fact that the most detailed data on eider flight height come from an offshore wind farm (excluded from the analysis) where these heights were relatively accurately measured by radar (Nysted, Denmark). All Arctic skuas and great skuas recorded by Cook et al. (2011) will have been migrating, as none of these species nest anywhere near any of the sites where their flight heights were recorded, yet their data were not excluded on the grounds of being migrants. Indeed, flight heights of skuas reported by Cook et al. (2011) are lower than the normal flight heights of skuas in areas around breeding colonies (R.W. Furness pers. obs., and preliminary data from GPS loggers deployed on great skuas by H. Wade, C. Thaxter and colleagues) suggesting that these species may fly lower over the sea when migrating than when foraging. Given the present locations of offshore wind farms around the UK, it is probable that a high proportion of many of the seabird species whose flight heights are reported in Cook et al. (2011) were migrants, and it would seem inappropriate to exclude such birds from assessment of collision risk. In cases where flight heights of seabirds have been measured by radar, the data can differ quite considerably from those reported by Cook et al. (2011) that are predominantly derived by observers viewing from boats. For example, Cook et al. (2011) report that 13% of black-headed gull flights are at blade height. However, radar studies reported a mean flight height of black-headed gulls of 29m (Day et al. 2003, Walls et al. 2004, Parnell et al. 2005), which implies a higher proportion at blade height. This discrepancy is unexplained, but it seems likely that the radar measurements of flight height are more reliable. It would be valuable to assess species risk directly from a much more complete data set on measured seabird flight heights at a variety of sites, and under different environmental conditions. We entirely agree with Cook et al. (2011) conclusion 'There is an urgent need for further research into the flight heights and avoidance rates of seabirds in relation to offshore wind farms. Ideally, this would include direct measurements of these variables through the tagging of individual birds and the monitoring of movements at a broader scale through the use of technologies such as radar'

Where data did not match up between Garthe and H◘ppop (2004) and Cook et al. (2011) (see Table 5), we estimated a value that appeared more consistent with data from closely similar species, or with other published data such as Rothery et al. (2009). For species not listed in either of the two reviews, data taken from the literature were used wherever possible to estimate an appropriate value ( Table 6). Although flight altitude is clearly an extremely important factor, other local aspects of siting can also be influential at specific sites. For example, turbines placed between a common tern colony and their feeding habitat have had a high impact on a particular colony (Everaert and Stienen 2007, Stienen et al. 2008), which might not be the case where a wind farm is placed away from a mandatory flight line of birds from a specific breeding site.

Table 6. Flight height estimates