Abstract
- Offshore windfarms are likely to become one of Europe's most extensive technical inventions in marine habitats. The areas in which windfarms are located often coincide with areas favored by large concentrations of seabirds. The UK Government has a legal obligation to monitor the impact windfarms might have on seabirds.
- This report follows a previous report commissioned by COWRIE (Maclean et al. 2006), in which the extent to which BERR (formally DTI) aerial survey protocol allows changes in bird numbers to be detected, was examined using power analysis. Changes in bird numbers can be hard to detect due to natural fluctuations in bird numbers. Power analysis allows a statistical assessment to be made of the probability that such changes could be distinguished from background fluctuations.
- Several issues are assessed in more detail than in the previous report. Rather than using data collected in many areas to estimate variability by assuming that the extent to which birds fluctuate in numbers is broadly similar (relative to the mean number present) across all sites, we select a few areas for which a long time-series of data exists and analyze these data exclusively. Additionally, rather than assuming a uniform decline within any given area, we explore one method of allowing for a gradient of decline, such that declines after windfarm construction are greatest closest to the windfarm. Lastly, we investigate the effects of using simple spatial variables (water depth, seabed slope gradient, seabed aspect, distance from land, distance from shallow water and seabed complexity) on the ability to detect changes in bird numbers. As with the previous report, four taxa were selected for analysis: red-throated diver (Gavia stellata), common scoter (Melanitta nigra), sandwich tern (Sterna sandvicensis) and lesser and great black-backed gull (Larus fuscus and L. marinus).
- Using long time-series of data rather than assuming equal proportional variability across many sites has no appreciable difference on the likelihood of being able to detect declines in bird numbers of 50% or less. The statistical power of being able to detect such changes remains low (
- Analysing data by assuming a gradient of decline generally results in a lower power to detect changes in numbers, but can sometimes improve power if declines are particularly severe. This is unsurprising given that the over-riding effect of using a gradient of decline, rather than uniform decline to estimate variability, is to increase the variability. Again, the statistical power of being able to detect such changes remains low (
- Similar conclusions are drawn in this report compared with the previous displacement report. The statistical power of being able to detect changes in bird numbers is lower than desirable. This is primarily because there are large inter-annual fluctuations in numbers. In our opinion, the only way in which changes in numbers could be detected with a high degree of certainty would be to find ways of explaining some of this temporal variation, such as through the incorporation of dynamic ocean variables into analysis. Failing this, we caution that an inability to detect changes in numbers power windfarm construction should not be taken to mean that no such changes are occurring.