Abstract
Offshore renewable developments (ORDs) have the potential to affect seabirds, through displacement from important at-sea foraging habitat. The estimation of displacement mortality rates is a critical component of the assessment process for estimating impacts on seabirds arising from ORDs, with large influence upon resulting predicted ORD impacts for affected populations.
Displacement mortality rates form a key input to assessment tools such as the Displacement Matrix, as well as comprising part of predicted outcomes from other assessment tools such as Individual Based Models (e.g., SeabORD, Searle et al. 2018). However, there is very little empirical evidence upon which these rates may be based. This project aimed to critically review the displacement mortality rates for seabirds used to determine the mortality of birds displaced by offshore windfarms in the UK offshore wind assessment process. And, where feasible, to provide updates to species-level estimates of mortality rates, with a clear indication of current supporting evidence and associated uncertainty. The project also sets out high-level recommendations for future empirical and analytical work addressing key knowledge gaps.
The project first reviewed existing information on the displacement mortality rates used to determine the mortality of birds displaced by ORDs in the UK. The review focused on six key species considered to be at greater potential risk of displacement and displacement mortality in future offshore windfarm development in the UK: Black-legged kittiwake Rissa tridactyla; Common guillemot Uria aalge; Razorbill Alca torda; Atlantic Puffin Fratercula arctica; Red-throated diver Gavia stellata; and Northern Gannet Morus bassanus. These reviews found no empirical evidence of the mortality of seabirds that have been displaced from offshore windfarms. It appears likely that this has not been studied to date, rather than not reported.
An Expert Elicitation workshop was then run to elicit mortality rates for these six species in the context of displacement from ORDs in both the breeding and non-breeding seasons. Given the uncertainty associated with displacement mortality in general, and for these species in particular, the workshop sought to address the knowledge gaps related to the excess mortality rates for these species, for both individual birds and their inter-dependents. The purpose of eliciting experts’ judgements regarding these knowledge gaps was to gain a better understanding of the biological context in which displacement mortality may occur, and to provide information of relevance to the implementation of current assessment methodologies (Joint-SNCB 2022). Broadly, the bulk of belief for the Displacement Mortality Rates (DMR) for individuals (breeding or non-breeding season) was on values below 10%. There was, however, notable variance in the estimated upper bounds, meaning disagreement and overall uncertainty in what would be a plausible upper limit to the effects of OWF in terms of DMR. In terms of the effects on dependents (chicks in the nest), there was markedly less agreement and certainty indicated by the experts’ responses. However, overall, the DMR for dependents was estimated as being substantially greater than for the mature individuals. Experts also discussed which aspects of seabirds’ interactions with offshore windfarms had not been considered in estimates, and identified key areas requiring future research:
• The importance of capturing displacement impacts on seabird productivity.
• Potential effects of habituation.
• That the values currently used in the assessments for the size of the populations that might be affected by DMR are noted as being under-estimates, being based on snapshots, perhaps limited to OWF footprints and near surrounds.
• Other classes of bird not considered in the EE may undergo differing displacement effects, including “Sabbatical” adults and juveniles.
• Other species were noted as being of potential importance.
• Currently the cumulative effect of multiple windfarms is done in a simplistic additive fashion, and more research is needed in how to better capture cumulative impacts from multiple ORDs.
An alternative way of quantifying displacement mortality rates is via simulation from a mechanistic model. SeabORD provides an individual-based model of movement, behaviour, energetics and demography in response to offshore renewables. SeabORD simulates impacts of displacement and barrier effects during the chick rearing period upon three key quantities: (a) chick mortality, (b) adult mass loss over the chickrearing period, and (c) over-winter adult mortality, which is derived from mass loss over the chick-rearing period. SeabORD is primarily designed to quantify population-level impacts, although it does this via simulation of individuals, in such a way as to incorporate inherent stochasticity. SeabORD does not use a “displacement mortality rate” as such, but such rates can be estimated using the model outputs. Complications arise because there are multiple possible ways to define such rates within the context of SeabORD. We consider two such definitions here: (a) rates that are defined in relation to the number of individuals displaced per timestep, and (b) rates are defined relative to the number of individuals displaced at any point during the chick rearing period. We ran SeabORD for three species (common guillemot Uria aalge, black-legged kittiwake Rissa tridactyla and razorbill Alca torda) at four SPAs, under a range of windfarm scenarios, to understand the relationship between the proportion of individuals that are displaced (per timestep, and over the entire season) and the impacts on mortality and mass loss. The slopes of these relationships, which we can estimate via statistical modelling (emulation), provide modelbased estimates of the displacement mortality rate. The emulation approach allows us to quantify uncertainty in these estimates, and to identify sources of variation in the rates. Where sources of variation are explainable in terms of colony and windfarm characteristics it also provides a basis for extrapolating displacement mortality rates to new scenarios, and thereby to estimating the population-level displacement risk under those scenarios. The results of the SeabORD runs and emulation suggest that the impacts on mortality and mass loss in SeabORD are relatively strongly driven by the proportion of individuals that are experiencing displacement, and there is no clear evidence for non-linearity in this relationship, when expressed in terms of the individuals experiencing displacement per timestep. However, the results suggest that the relationships can vary quite substantially between SPAs - more SeabORD runs would be required to understand the cause of this variation, and thereby to produce a version of the emulator that can infer the rates to use for additional SPAs. When the proportion of individuals experiencing displacement is relatively low the results are also relatively noisy – the SeabORD runs considered here only spanned higher levels of displacement for timestep for black-legged kittiwake, so for common guillemot and razorbill more runs, covering scenarios with higher levels of encounter with windfarms, would also be needed in order to reliably infer rates.
The displacement mortality rates generated via expert elicitation and SeabORD/emulation cannot be used within the Displacement Matrix, because of the differences in definition between the different approaches. In particular, the Displacement Matrix uses seasonal peak estimates of abundance derived from at-sea survey data to quantify the number of individuals at risk of displacement, which is difficult to relate to the definitions of the number of individuals at risk used in the EE and SeabORD/emulation. Further research on site fidelity and turnover is needed in order to understand these differences.
Our belief is that it is not appropriate to use the displacement mortality rates arising from the EE within the Displacement Mortality in its current form unless adjustments are made to account for this discrepancy. Our key recommendations are therefore for further work to understand site fidelity, and the impact of this upon displacement risk, to understand whether there is a potential for underestimation of effects within the Displacement Matrix.
This work would involve:
• Interrogation of GPS tracking data to estimate rates of fidelity in seabird species, including influence of environmental variation and seasonal variation.
• Examination of seabird time-activity budgets to understand influence of partitioning of behaviour between at-sea and colony behaviours and how this might be used to adjust at-sea survey data.
• Tracking of individual birds to link observed interactions with operational offshore windfarms (barrier effects and displacement) with subsequent demographic rates (breeding success and survival).