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

1. Introduction

Offshore wind farm developments play a key role in strategies to reduce our reliance on energy generated using fossil fuels (Toke 2011). However, there are also concerns about the potential for offshore wind farms to negatively impact the environment, with the risk to seabirds receiving particular attention (Furness et al. 2013). Consequently, Environmental Impact Assessments (EIAs) must be carried out as part of the consenting process for proposed offshore wind farms. In relation to seabird populations, the key potential effects are perceived to be collisions with turbines, the loss of habitat as a result of displacement and barrier effects resulting in elevated energy expenditure costs.

Collision risk is assessed as part of the EIA process using a Collision Risk Model (CRM) such as the Band model ( Band et al., 2007). The Band (2007) model was originally developed for use onshore, but refined in order to better reflect data collection methodologies in the offshore environment ( Band, 2012). In common with most CRMs, the Band (2012) model requires reliable estimates of behavioural parameters, including estimates of species-specific flight heights, flight speeds and levels of nocturnal activity (Masden & Cook, 2016). However, these parameters are subject to significant uncertainty and variability, and how this is accounted for within the Band (2012) model has not always been clear. Consequently, Masden (2015) updated the Band (2012) model to incorporate stochasticity by enabling users to input parameters as distributions rather than point estimates. This was subsequently updated and made available as a more user-friendly web application (McGregor et al., 2018).

At present, estimates of species flight heights have typically been based on boat or digital aerial survey data (Johnston & Cook, 2016; Johnston et al. 2014). Similarly, estimates of other key parameters such as flight speed and levels of nocturnal activity have been drawn from studies with limited sample sizes or have been based on reviews inferred from our understanding of the ecology of the species concerned (Alerstam et al. 2007; Garthe & Huppop, 2004). However, ongoing analyses have highlighted how using data in this way may give a mis-leading impression of collision risk (Masden et al., 2021). Analysis of Lesser Black-backed Gull flight heights has shown that they differ between day and night (Ross-Smith et al., 2016). Other recent analysis of bird flight speeds and nocturnal activity also reveal disparities with the generic values incorporated in collision risk models (Fijn & Gyimesi, 2018; Furness et al., 2018). These parameters may also be influenced by other factors including wind speed and direction, time of day and distance from the breeding colony (Thaxter et al., 2019). Given the sensitivity of collision risk models to these parameters (Masden, 2015), getting better estimates for these parameters and an improved understanding of factors influencing their spatial and temporal variation will be a key step for reducing the uncertainty associated with assessments of collision risk.

We collate GPS tracking data collected from lesser black-backed gull Larus fuscus, kittiwake Rissa tridactyla and gannet Morus bassanus and analyse these to generate estimates of flight height, flight speed and nocturnal activity for these species. Drawing on previous analyses (Thaxter et al., 2019), we investigate how flight height and speed vary between foraging and commuting behaviours. For lesser black-backed gull we also consider how levels of nocturnal activity vary over the course of the breeding season. Finally, we consider the implications of our results for the assessment of collision risk.