Abstract
Typically, ecological impact assessments (EIA) are conducted under time constraints, making the collection of baseline data and application of Before-After-Control-Impact (BACI) designs difficult. Here we report the results of three “post-development” experiments testing the effects of a small wind park on the abundance, distribution, and behavior of wintering Common Eiders (Somateria mollissima), a large sea duck. Our approach was based on the rationale that the probability of detecting an impact should increase with decreasing distances from the wind park. Because prey abundance is likely to drive the distribution of wintering eiders, we removed that confounding variable by (1) randomizing its effect over the study area, and (2) incorporating the variable into the analysis. In the first experiment, we compared the abundance and distribution of eiders when wind turbines were switched on and off. This was complemented by investigating the escape behavior of flocks when wind turbines were switched on after they had been off for one full day. In a second experiment, we tested for any departure from the habitat-matching rule, which states that the proportion of eiders and the proportion of food should be equal at different distances from the park. In a third experiment, the flying behavior of eiders in relation to the wind park was analyzed by measuring the landing and flying rates for patches of decoys located at 100, 300, and 500 m from the wind park. We found little evidence for negative impacts, because we could not detect any effect in three tests out of four. Only in the decoy experiment did we observe eiders reducing their landing and flying near (100 m from) the wind park. We conclude, for the conditions under which our experiments were performed, that the wind park did not substantially affect wintering Common Eiders. However, many aspects of the potential impacts of offshore wind parks on sea ducks have not been covered by our study and, therefore, cannot be generalized to other species or other phases of the annual cycle. Nevertheless, we argue that the use of multiple post-development experiments based on a gradient approach is a helpful complement to BACI studies.