Modelling flight heights of lesser black‐backed gulls and great skuas from GPS: a Bayesian approach

Abstract

1.Wind energy generation is increasing globally, and associated environmental impacts must be considered. The risk of seabirds colliding with offshore wind turbines is influenced by flight height, and flight height data usually come from observers on boats, making estimates in daylight in fine weather. GPS tracking provides an alternative and generates flight height information in a range of conditions, but the raw data have associated error. 2. Here, we present a novel analytical solution for accommodating GPS error. We use Bayesian state-space models to describe the flight height distributions and the error in altitude measured by GPS for lesser black-backed gulls and great skuas, tracked throughout the breeding season. We also examine how location and light levels influence flight height. 3. Lesser black-backed gulls flew lower by night than by day, indicating that this species would be less likely to encounter turbine blades at night, when birds' ability to detect and avoid them might be reduced. Gulls flew highest over land and lowest near the coast. For great skuas, no significant relationships were found between flight height, time of day and location. 4. We consider four 'collision risk windows', corresponding to the airspace swept by rotor blades for different offshore wind turbine designs. We found the highest proportion of birds at risk for a 22-250 m turbine (up to 9% for great skuas and 34% for lesser black-backed gulls) and the lowest for a 30-258 m turbine. Our results suggest lesser black-backed gulls are at greater risk of collision than great skuas, especially by day. 5. Synthesis and applications. Our novel modelling approach is an effective way of resolving the error associated with GPS tracking data. We demonstrate its use on GPS measurements of altitude, generating important information on how breeding seabirds use their environment. This approach and the associated data also provide information to improve avian collision risk assessments for offshore wind farms. Our modelling approach could be applied to other GPS data sets to help manage the ecological needs of seabirds and other species at a time when the pressures on the marine environment are growing.

Atmospheric conditions create freeways, detours and tailbacks for migrating birds

“…There seems to be a general consensus that flapping flight of seabirds at sea, though not necessarily during migration, occurs at relatively low flight altitudes <100 m (Garthe and Hüppop 2004; Ross-Smith et al 2016). Within this relatively narrow altitude band, studies have shown that as expected from optimal migration theory seabirds as well as waterbirds fly at higher altitudes with supporting winds and lower altitudes with opposing winds (Krüger and Garthe 2001; Kahlert et al 2012; McLaren et al 2016) with waterbirds also reducing their flight altitude with increasing wind speed (Kahlert et al 2012).…”

State‐space modelling of the flight behaviour of a soaring bird provides new insights to migratory strategies

“…Beyond energetics, characterising behavioural states in flying animals is particularly important to evaluate their susceptibility to human activities, informing effective planning and management (Katzner, Brandes et al., ; Péron et al., ; Ross‐Smith et al., ). For example, specific behavioural states, due to their horizontal and vertical characteristics, may put birds at higher risk of collision with turbines (Ross‐Smith et al., ).…”

Improving estimation of flight altitude in wildlife telemetry studies

“…Ross‐Smith et al. () incorporated DOP as a linear effect into state‐space models showing that variation in DOP was strongly related to variation in flight altitude error. That error is also relevant when considering Péron et al.…”

“…Finally, Ross‐Smith et al. () developed similar models that incorporated negative observations, but that ensured that all estimated true altitude values were above sea level (ASL; positive altitudes). Although the incorporation of flight altitude values into state‐space models is more complex than the use of a threshold, such models have their own weaknesses that may limit their utility and may affect ecological inference (Auger‐Méthé et al., ).…”

“…This has been done in a number of different ways in the past. Recently, state‐space models have been fitted to flight altitude data collected from GPS telemetry units placed on birds (Péron et al., ; Ross‐Smith et al., ). These models separate the variance in flight altitude data into two components described as process (actual vertical movements) error and sampling (measurement or observation) error, and then adjust the flight altitude values by correcting for the measurement error.…”