Developing a population model for Rum Manx Shearwaters for assessing offshore wind farm impacts and conservation measures

2 Introduction

The Manx shearwater (Puffinus puffinus) population situated in the mountains of Rum is of international importance. Given proposed offshore wind farm (OWF) developments in the north-west region of Scotland’s seas (Marine Directorate 2020), this population may be at potential risk from displacement/barrier effects on adults (Searle et al. 2018) or collision mortality of naïve fledglings attracted to the safety lights used by wind farms to warn aircraft (Syposz 2020). Additional threats are posed to breeding success from a population of invasive brown rats Rattus norvegicus (Burnell et al. 2023) and changing climate, extreme weather events (e.g., resulting in flooding of burrows). It is therefore important to establish a baseline for this population, to evaluate risk with as much precision as the available data allow and to produce predictions of how the Rum population may respond to additional anthropogenic mortality or natural disturbance. The global breeding population is estimated at between 744,700 and 1,222,500 pairs of which the colonies in Britain and Ireland hold between 712,010 and 1,184,800 pairs or some 96% of the total (Burnell et al. 2023). It can therefore be considered as the most endemic of all seabirds breeding in these islands. There are relatively few colonies (50 in total in Britain and Ireland) but 80% of the population is concentrated in Pembrokeshire (Skokholm and Skomer) and on Rum, with some uncertainties around population estimates at all three colonies (Burnell et al. 2023). In particular, the most recent estimate for the Rum colony (Inger et al. 2022) is substantially larger than previous estimates, as is discussed in detail later in this report. Nine islands off Kerry in SW Ireland hold a further c.14% of the British and Irish population. There is limited information on metapopulation structure, but it is expected that some transfer of young birds to other colonies occurs. Adult breeders normally return to the same burrow each year if breeding is successful but may move to a nearby burrow if they fail.

There is currently little overlap between Manx shearwater at-sea distribution and existing offshore wind farms (installations in the Irish Sea are unlikely to be an important area for birds from Rum, although birds from Rum do use the Irish Sea front area (Guilford et al. 2019). However, some of the lease areas of the ScotWind programme will see developments in areas used by Manx shearwater. Possible displacement/barrier effects are always a risk (Searle et al. 2018) but another concern (Syposz 2020) is that fledgling shearwaters can be attracted to the safety lights used by wind farms to warn aircraft and wider interactions with lighting at sea are uncertain (Deakin et al. 2022). It is therefore important to develop a baseline for this population, to evaluate risk with as much precision as the current aggregate data allows and to produce a population model that generates predictions of how the Rum population may respond to additional mortality/reduced breeding success arising from anthropogenic factors.

This project aimed to apply Bayesian state-space modelling to relevant available data including surveys of the Rum population, long term but discontinuous monitoring of burrow occupancy and breeding success in small sample areas, ringing records, weather data and previous studies of rat depredation and rainfall effects. The overarching scientific objective was to construct a robust model of the dynamics and drivers of the Rum shearwater population. Work package references can be put into context by referring to the points below.

Work structure for the entire project, included here for reference.

• Data collation: Formatted data frame of available survey, demographic and covariate data. Effort to provide precision for each datum. Combine information and expert opinion into formal parameter priors.
• Model development: Biological and observation process models. Validation of model explanatory and predictive performance with realistic simulated data.
• Covariate exploration: Exploring the strength of various covariates and association with individual demographic rates.
• Hindcasts and forecasts: Reconstructing past trajectories for population size and three demographic rates. Forecasts and associated credible intervals, including scenarios for avian flu and windfarm effect.
• Expandability: A review and report on the framework’s scalability, expandability and compatibility.