Offshore wind developments assessment - seabird collision risk, displacement and barrier effects: study

6 Discussion

Here, we first summarise outcomes from an initial project workshop to explore approaches for combining displacement, barrier and collision assessment approaches. We then identify a series of future research priorities for integrating modelling methods for collision and displacement/barrier impacts. Finally, we present a set of wider recommendations to facilitate and advance approaches for combined estimates for collision and displacement/barrier impacts on seabirds.

6.1 Workshop on combining displacement, barrier and collision assessment methodologies

The project held an initial workshop that brought together experts from research, environmental consultants, government and conservation bodies to discuss how best to implement combined modelling of displacement, barrier and collision risks from OWFs for breeding seabirds. The aim was to discuss how current methods are used to estimate collision and displacement risks separately in assessments, with a following session on the extent to which the inputs, parameters and assumptions used in the different methods are consistent with each other. Finally, there was a session on how to combine collision and displacement risks into a single assessment of risk. For full reporting of workshop discussions and conclusions see Appendix C. The main recommendations arising from the workshop were as follows:

6.1.1 General recommendations from workshop

By bringing together experts in collision risk and displacement modelling, we were able to summarise the key challenges in integrating collision risk and displacement, and agree an approach for integrating sCRM outputs (collision) into SeabORD or with the matrix method (displacement and barrier) in this project, using black-legged kittiwake (to illustrate practical combining of the sCRM with the individual-based model, SeabORD), and a more general discussion for combining the sCRM + matrix approaches

In using GPS tracking data there is a need to integrate outputs from hidden Markov models (HMMs) by BTO to partition movement data by behaviour (Thaxter et al. 2019), thereby allowing behaviour-specific bird utilisation distributions to be used within models;

Flight height should consider commuting and foraging flight separately when applied to an individual-based model; for some species (e.g. black-legged kittiwake) flight speeds should also be estimated and modelled separately for commuting and foraging; if birds use distinct areas for commuting and foraging, then using an average flight height may overestimate collision risk in commuting areas and under-estimate collision risk in foraging areas (assuming birds fly higher when foraging) (Cleasby et al. 2015). Given different locations/size/layouts etc. of wind farms, it's probably not reasonable to assume the two biases would cancel each other out.

There is a need to incorporate 3D flight movements, based on the latest GPS tracking data (available for gannet and kittiwake, more planned for both species at more locations; e.g. Thaxter et al. 2017);

There is a need for further empirical data and validation of inputs and outputs of sCRM, matrix approach and SeabORD;

Further model developments must consider the growing relevance of cumulative in-combination effects across multiple OWFs. Some progress has been made but more will be needed.

6.1.2 Specific sCRM recommendations

There is a need to parallelise sCRM to increase its speed to levels obtained when using the underlying code;

There is a need to incorporate an observed distribution of flight speeds in sCRM (a truncated normal distribution with a mean and standard deviation is currently used);

6.1.3 Specific SeabORD recommendations

There is a need to speed up running time for model, and include automated calibration to handle application with new data more efficiently.

6.2 Future research priorities for integrated modelling of displacement and collision

6.2.1 Unifying models and inputs

We have developed a framework for simultaneously estimating collision and displacement effects, from either a single OWF or multiple OWFs, using an individual-based mechanistic model of seabird movement, foraging and demographics coupled with output from the sCRM. In the initial implementation of this framework, we have used outputs from the existing sCRM model to provide information on collision-related mortality rates, which are then used to simulate collision events within an extended version of SeabORD.

This integration has highlighted some key complexities around some of the model parameters, notably the need for methods to empirically estimate macro-avoidance to improve the realism of parameters used within the new combined model and consistency in assumed rates of displacement. A further key issue for future work is to develop methods which would enable GPS data to be used to estimate flux within collision risk models.

It would ultimately be preferable to unify the two models into a single model, by embedding the sCRM calculations within SeabORD so that the Band model calculations of collision risk would effectively be performed for each bird on each day. Unifying the models would improve usability and reduce the potential for users to run the two models in ways that are inconsistent with each other (e.g. by providing inputs with inconsistent values).

The key advantage of unifying the models using an individual-based modelling approach, however, would be the potential to refine the model processes to improve the realism of the biological assumptions of both models. This would include, for instance, changes to simulate 3D flight paths for individual birds within OWFs, and more realistic (non-straight line) foraging paths outside of OWFs, as well as potentially separating out collision risk into specific behaviours, such as commuting flight and foraging flight. The key advantage in modelling individual flights within OWFs is that the model would automatically separate out the three types of avoidance – macro (displacement/barrier), meso (>10 m movements to avoid individual turbines), and micro (<10 m 'last second' flight adjustments close to individual turbines). It would also be desirable to extend the SeabORD model to include the incubation period, as well as the chick-rearing period within model simulations.

However, it is important to note that not all species need both collision and displacement to be modelled simultaneously. For some species, statutory advice has so far been that collision risk modelling is not needed, primarily due to the flight height distribution of birds (e.g. shearwaters, auks), while for some species collision risk is important, but displacement appears not to occur (e.g. large gulls or other species with very large foraging ranges such as northern fulmar and northern gannet, in Scotland only). An integrated model will be of most value to those species where the likely impacts from displacement and/or barrier effects and collision risk are both considered important (e.g. black-legged kittiwake and northern gannet). In addition, stand-alone collision risk modelling will continue to be useful for determining the worst-case scenario within the design envelope of the proposed OWF.

6.2.2 Better methods for incorporating non-breeding birds at risk of collision and displacement within modelling approaches, and new research needed to extend modelling to include immature and sabbatical birds (non-breeders), and to potentially deal with birds breeding in other countries

New science is needed to extend individual-based models such as SeabORD to cover the whole of the breeding season (but see van Kooten et al 2019 for initial work in this area). At present, for each of the four species currently parameterised within SeabORD, the model only simulates OWF impacts over the chick-rearing period. Currently, a full extension to the whole of the breeding season (pre-breeding attendance, incubation, chick-rearing, post-fledging attendance) may not be possible for all seabird species at risk of OWF impacts due to a lack of data on individual movements, behaviour and other ecological processes. This development would require several key stages: 1. Identification, collation and processing of relevant data, 2. Theoretical model development to incorporate new behaviours and processes out-with the chick-rearing period, 3. Implementation of new developments within the model code and subsequent testing, and 4. Model validation, QA and sensitivity analysis.

In order to fully reflect ecological reality, it is important that impacts on all life-history stages are considered, including those on non-breeding birds (e.g. immature birds and those taking a sabbatical). Estimates of the proportion of immature birds within the population could be obtained by examining digital aerial images (for species where plumage differs recognisably between age classes), while estimates of the proportion of birds taking a sabbatical year could be determined through the analysis of ringing and colour-ringing datasets where available. In order to better apportion impacts back to the appropriate protected sites, a clearer understanding of the movements of birds between colonies, both in the UK and elsewhere, is needed (Black & Ruffino 2018; Ruffino et al. 2020). Previous large scale-analysis of GPS tracking data has highlighted the partitioning of birds from different colonies at sea (e.g. Wakefield et al. 2014). Given the rapid expansion of GPS tracking studies, similar analyses should be considered for other species in order to gain a clearer understanding of how birds from different colonies may interact with particular developments.

Other methods for estimating impacts on non-breeding birds should also be explored, such as more empirical work aimed at quantifying the proportion of non-breeders associated with SPA populations, movements between SPA colonies, and the relative abundance of non-breeders or birds from non-UK colonies observed at sea.

There are important differences in the approaches needed for undertaking impact assessments for different legislation. EIA legislation is intended to protect the environment as a whole, while Habitats Regulations Appraisal (HRA) is intended to protect specific populations that utilise and are supported by a network of designated sites. Thus, EIA needs to apply to whole populations of seabirds at regional and national/international levels and includes all birds regardless of their age or breeding status. HRA needs to apply only to impacts that affect the conservation objectives of the site. Overarching objectives tend to focus on maintaining the designated population size. In the case of SPA designated seabird colonies those populations are breeding adults. While it is important to consider impacts on other demographic elements (e.g. sabbatical birds, immature birds, etc.) it is only where these are relevant to maintain the designated population size of adult birds that they need to be included in the impact assessment. Thus, current individual based models are much more useful to the HRA than to the EIA, as the population is well defined and individuals that are part of the qualifying population of a site are a suitable unit of assessment. HRA is also intended to be a higher hurdle to development with the potential to do harm, reflecting the greater conservation importance on species and populations within designated sites.