Understanding seabird behaviour at sea part 2: improved estimates of collision risk model parameters

Author(s): Cook, A.S.C.P., Thaxter, C.B., Davies, J., Green, R.M.W., Wischnewski, S. & Boersch-Supan, P.

Published: October 2023  

ISBN: 978-1-83521-271-4

View publication


Collision risk models form a key part of the environmental impact assessments for offshore wind farms. These models require inputs on species behaviour including flight speed, flight heights and nocturnal activity rates. At present, these are largely based on generic estimates from the literature, some of which are largely qualitative rather than quantitative. However, the rapid expansion of GPS tracking offers a valuable source of data for information on these parameters.

We collated GPS data from gannet, kittiwake and lesser black-backed gull, three species of key concern in relation to collision risk, from across multiple colonies. As a first step, we use Expectation-Maximisation Binary Clustering (EMbC) to classify GPS tracks as relating to foraging or commuting behaviour. These data were used to estimate the proportion of time birds were active during the night (e.g., classified as foraging or commuting) relative to the proportion of time birds were active during the day in order to quantify nocturnal activity rates. This parameter is difficult to estimate using other methods for collecting collision risk model parameter information, e.g. visual records. For all three species, nocturnal activity levels (e.g., the proportion of time birds were active at night, relative to the proportion of time birds were active during the day) varied between sites and years. However, the proportion of time birds were active during the night remained relatively constant, suggesting that this variation was driven by activity levels during the day.

Flight speeds were estimated separately for foraging and commuting flight. We used two different approaches to estimate flight speed – instantaneous speeds measured directly by the GPS tags using the Doppler effect, and trajectory speeds based on the distance travelled between fixes. In both cases, data were highly autocorrelated and had to be sub-sampled to one point every hour. Trajectory and instantaneous speeds were highly correlated with one another; however, the trajectory speeds were consistently an underestimate, compared with instantaneous speeds, an effect which became more pronounced as the sampling interval of the tags increased. Regardless of the approach used, foraging flight speeds were consistently lower than commuting flight speeds, and all speeds were lower than the current generic estimates of flight speeds recommended for use in the Band Collision Risk Model.

Flight heights were also estimated separately for foraging and commuting flight. Of the species considered, lesser black-backed gulls typically flew higher, and spent a greater proportion of time at collision risk height, than was the case for either gannets or kittiwakes. Commuting flights typically took place at greater heights than foraging flights. For lesser black-backed gulls, whilst there was consistency in foraging flight heights, there was evidence of differences in commuting flight heights among colonies.
GPS tracking provides a valuable source of data for parameterising collision risk models. However, there are clear differences in flight heights and speeds between areas used for foraging and commuting behaviour, which are likely to have implications for collision risk. The greater heights and faster speeds of commuting birds are likely to lead to greater estimates of collision risk than is the case for foraging birds. Furthermore, colony-specific differences in both nocturnal activity rates and flight heights suggest that some degree of site-specific data may be required for reliable parameterisation of collision risk models in ornithological impact assessments

Staff Author(s)

Related content