A corrigendum was published by the authors on 6 June 2023
Decline of Leach’s Storm Petrels Hydrobates leucorhous at the largest colonies in the northeast Atlantic
* Correspondence author. Email: firstname.lastname@example.org
1School of Biosciences, Cardiff University, Museum Avenue, Cardiff CF10 3AX, UK;
2South Iceland Nature Research Centre, gisgata 2, 900 Vestmannaeyjar, Iceland;
3National Trust for Scotland, Hermiston Quay, Cultins Road, Edinburgh, EH11 4DF, UK;
4School of Natural and Social Sciences, University of Gloucestershire, Cheltenham, GL50 4AZ, UK;
5Department of Zoology, Oxford University, Oxford OX1 3SZ, UK;
6Cardiff School of Dentistry, Cardiff University, Heath Park, Cardiff CF14 4XY, UK;
7RSPB, The Lodge, Sandy, Bedfordshire SG19 2DL, UK;
8RSPB Centre for Conservation Science, The Lodge, Sandy, Bedfordshire, SG19 2DL, UK;
9RSPB Northern Ireland, Belvoir Park Forest, Belfast BT8 7QT, UK;
10NatureScot, Achantoul, Aviemore, Inverness-shire PH22 1QD, UK;
11The Open Seas Trust, 51 Atholl Rd, Pitlochry PH16 5BU, UK.
Leach’s Storm Petrel Hydrobates leucorhous is a widespread and highly pelagic seabird, breeding in burrows on islands across the Atlantic and Pacific Oceans. The global population is estimated at 6.7–8.3 million breeding pairs, but sharp declines have been detected at Atlantic colonies, leading to the species being up-listed from ‘Least Concern’ to ‘Vulnerable’ on the IUCN Red List in 2016 (BirdLife International 2020). The main eastern Atlantic Leach’s Storm Petrel colonies are in Iceland and Scotland (Mitchell et al. 2004), but there are also up to 1,000 pairs breeding in Mykineshólmur in the Faroe Islands and smaller numbers breeding in Norway and Ireland (Bolton & Eaton 2020).
The Vestmannaeyjar archipelago contains almost all of Iceland’s breeding Leach’s Storm Petrels and, based on extrapolation of the densities measured on Elliðaey Island in 1991, is believed to hold the largest population in the eastern Atlantic (Hansen et al. 2009). The 1991 survey of Elliðaey revealed a strong positive association between the occurrence of Leach’s Storm Petrel burrows and Atlantic Puffin Fratercula arctica (hereafter ‘Puffin’) habitat, which had been mapped across the whole of the Vestmannaeyjar archipelago (Hansen et al. 2009). Six other islands in the archipelago hold breeding colonies of Leach’s Storm Petrels: Bjarnarey, Álsey, and Suðurey, which contain similar-sized areas of breeding Puffins to Elliðaey, and Brandur, Hellisey and Smáeyjar, which contain smaller areas of breeding Puffins (Hansen et al. 2011). The density of Leach’s Storm Petrel Apparently Occupied Sites (AOS; i.e. estimated breeding pairs) within surveyed areas of Puffin habitat on Elliðaey was extrapolated across the Puffin habitat on the other islands in the archipelago, to produce a whole-archipelago estimate of 178,900 (± 34,100) AOS, including 44,100 (± 9,100) AOS for Elliðaey Island (Hansen et al. 2009).
The Seabird 2000 census (Mitchell et al. 2004) included the first attempt to produce accurate population estimates for Leach’s Storm Petrels in Britain and Ireland. Mitchell et al. (2004) estimated the total British and Irish population of Leach’s Storm Petrels to be 48,357 AOS (95% CI: 36,742–65,193), with 94% of these in the St Kilda archipelago. The largest sub-colony on St Kilda was on the island of Dùn, with an estimated 27,704 AOS (95% CI: 20,430–38,506) in 1999/2000 (Mitchell et al. 2004). Further surveys of Dùn in 2003 and 2006 produced estimates of 14,490 (95% CI: 12,110–17,439) and 12,770 (10,046– 17,086) AOS respectively, suggesting a decline of 54% on the island since the Seabird 2000 survey (Newson et al. 2008).
Previous censuses of Leach’s Storm Petrels in the northeast Atlantic have used the acoustic playback method described in Gilbert et al. (1998), which is based on techniques developed for surveying European Storm Petrels H. pelagicus (Ratcliffe et al. 1998). The playback method involves playing recordings of storm petrel calls, which elicit responses from birds in nest sites. Not all storm petrels will respond to playback on every occasion, so the number of responses obtained during a survey is lower than the actual number of AOS in the surveyed area (Ratcliffe et al. 1998). Traditionally, storm petrel playback censuses use a multistage method. Playback is performed within a calibration plot, in which the actual number of AOS is known or estimated, in order to estimate the response rate, and thus obtain a correction factor. The total population size is then estimated by applying the correction factor to the number of responses detected in the main survey across a much wider area. Various analytical methods have been used to estimate response rates from the calibration plot (Mitchell et al. 2004; Bolton et al. 2010), which all rely on the assumption that there is an equal probability of response from all nests in the calibration area. Typically, the calibration area is divided into very small sub-plots (quadrats), with playback conducted in each, to satisfy this condition. Here, we term this approach the ‘calibration plot method’.
Traditional acoustic playback methods for burrow-nesting seabirds are extremely time consuming. Detected response rate declines with increasing distance of AOS from the playback point (Ratcliffe et al. 1998) so, to maximise response rates, survey quadrats are typically very small (1–4 m2). The ability to survey a larger area during each playback event, for example by using distance sampling methods which explicitly model the distance-detection function (i.e. the decline in detection probability as distance from the observer increases), has the potential to reduce the survey effort required to estimate a population size.
A key assumption of traditional distance sampling is that perfect detection occurs at distance = 0 (Buckland et al. 2001). However, the response rate of burrow-nesting seabirds to playback at distance = 0 tends to be substantially less than one, so there is a need to modify the method for these species. Hierarchical distance sampling (HDS) is a development of the distance sampling method that relaxes this assumption, by using repeat surveys of the same points to independently estimate (i) population density, (ii) detection probability and (iii) availability for detection (Sillett et al. 2012; Kéry & Royle 2016). The HDS method assumes that the population is closed, so all individuals are always present within the survey area, but allows for individuals to be unavailable for detection on some occasions, for example, if individuals move to unobservable locations at certain times of the day, or under particular weather conditions. For storm petrels, the probability of responding to playback can be treated as availability for detection, since it is not possible to detect birds when they do not respond. In traditional playback methods, the estimated ‘response rate’ is equivalent to the product of response rate (the probability of a bird responding to playback) and detection rate (the probability of the observer hearing an emitted response), but the HDS method estimates these components separately. As with the calibration plot method, HDS requires at least some points to be surveyed on more than one occasion, but it does not require individual AOS to be marked, as is the case with the calibration plot.
An additional advantage of HDS is that the density of birds can be modelled with respect to fine scale environmental covariates relating to habitat type. Although the calibration plot method can be used to estimate different densities in different habitat types, these tend to refer to broad areas, in which density is assumed to be homogeneous. A significant drawback of the calibration plot approach is that the colony area may be very difficult to delineate accurately, and errors in the assessment of colony area can hugely influence the resulting population estimate. This also applies to HDS to a degree, but an advantage of HDS is that models can incorporate covariates which explicitly represent the suitability of the habitat for the focal species.
In this study, our primary aim was to estimate current size and change of the largest Leach’s Storm Petrel populations in the northeast Atlantic. We present the results of a 2017–18 survey of Leach’s Storm Petrels on Elliðaey, and a 2019 survey of Leach’s Storm Petrels in the St Kilda archipelago; the latter conducted as part of the fourth national breeding seabird census of Britain and Ireland, ‘Seabirds Count’. At both colonies we used two playback survey methods, in an attempt to optimise the accuracy and precision of estimates in the time available, while also enabling back-compatibility and direct comparisons with previous surveys. On Elliðaey, two transects were surveyed in 1991, covering approximately 1% of the island’s area. We repeated these transects in 2017 and 2018 and also analysed data from a whole-island census based on a grid of sample points, using HDS. We used insights from these analyses to re-evaluate likely population size in 1991. For St Kilda, we used HDS but also analysed the survey data using the calibration plot method that was used in the previous censuses. We also evaluated the field and analysis methods in terms of their efficiency of data collection and precision of the resulting population estimates.
The expert boat handling skills of Angus and Alexander Campbell of Kilda Cruises were invaluable in achieving safe landings on the St Kilda islands. We thank the St Kilda Rangers, Sue Loughran, Sarah Lawrence and Craig Stanford for their support, Ian Stevenson of the Soay Sheep Project for providing weather data for St Kilda, and Historic Environment Scotland for providing LiDAR data for the archipelago. Work on Leach’s Storm Petrels on St Kilda was carried out under a Schedule 1 Disturbance Licence. SINRC provided transport to Elliðaey Island. Many thanks to Mark Day for initiating the fieldwork in Iceland, and to Steffen Oppel for suggesting the use of hierarchical distance sampling models and advising on their construction. Thanks to the two anonymous reviewers and to April Hedd for helpful comments on an earlier version of the manuscript. ZD is supported by a NERC GW4+ Doctoral Training Partnership studentship from the Natural Environment Research Council [NE/L002434/1] and is grateful to Frank Hailer for his support.
The Leach’s petrel survey of the St Kilda archipelago is part of ‘Seabirds Count’ (2015–21), the fourth national census of the UK’s breeding seabird populations. It was made possible using funding received by RSPB and JNCC from EDF Renewables, Moray Offshore Windfarm (West) Limited (Moray West), Red Rock Power Limited (Red Rock Power), and SSE Renewables.
IUCN Red List for birds. www.birdlife.org. Accessed 20 July 2020.
Ecological Models and Data in R. Princeton University Press, Princeton.
Emdbook: Ecological Models and Data in R. R package version 1.3.12.
An Introduction to Distance Sampling. Oxford University Press, Oxford.
Boot: Bootstrap R (S-Plus) Functions. R package version 1.3-22.
The Skuas. Poyser, London.
Bird Monitoring Methods: A Manual of Techniques for Key U.K. Species. Royal Society for the Protection of Birds, Sandy.
The Leach’s Storm-petrel population size of the Westman Islands. In: Stienen, E., Ratcliffe, N., Seys, J., Tack, J., Mees, J. & Dobbelaere, I. (eds.) Seabird Group 10th International Conference. Provincial Court, Brugge, 27–30 March 2009. VLIZ Special Publication 42. Communications of the Research Institute for Nature and Forest - INBO.M.2009.1.
The breeding population size of Atlantic Puffin in Vestmannaeyjar, S-Iceland. Bliki 31: 15–24.
Puffins on St Kilda. British Birds 70: 50–65.
Birds of St Kilda. Institute of Terrestrial Ecology, Cambridge.
caret: Classification and Regression Training. R package version 6.0-84. CRAN.R-project.org/package=caret.
Seabird and Marine Ranger Annual Report St Kilda 2019. National Trust for Scotland Report.
Survey of Atlantic Puffin breeding colonies on St Kilda in 2018. National Trust for Scotland Report.
Ecology, behaviour and predator-prey interactions of Great Skuas and Leach’s Storm-petrels at St Kilda. PhD thesis, University of Glasgow, UK.
A survey of Leach’s Petrels on Shetland in 2011. Scottish Birds 32: 22–29.
Seabird Populations of Britain and Ireland: Results of the Seabird 2000 census (1998–2002). Poyser, London.
The Status of Breeding Seabirds on St Kilda in 1999 and 2000. Unpublished JNCC report.
Classes and methods for spatial data in R. R News 5, 9–13. cran.r-project.org/doc/Rnews
Migration routes and stopover areas of Leach’s Storm-Petrels Oceanodroma leucorhoa. Marine Ornithology 47: 55–65.
ArcticDEM. doi.org/10.7910/DVN/OHHUKH, Harvard Dataverse, V1.
R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. www.R-project.org.
High apparent survival of Leach’s Storm-petrels Oceanodroma leucorhoa in British Columbia. Marine Ornithology 48: 133–140.
Herring Gull Larus argentatus predation on Leach’s Storm-petrels Oceanodroma leucorhoa breeding on Great Island, Newfoundland. Atlantic Seabirds 2: 35–44.
Sexual Dimorphism of Chatter-calls and Vocal Sex Recognition in Leach’s Storm-Petrels (Oceanodroma leucorhoa). The Auk 106: 498–501.
Update and trends of three important seabird populations in the western North Atlantic using a geographic information system approach. Marine Ornithology 43: 211–222.