Skip to main content

Use of sea ice by arctic terns Sterna paradisaea in Antarctica and impacts of climate change

Cite this dataset

Redfern, Chris P. F.; Bevan, Richard M. (2019). Use of sea ice by arctic terns Sterna paradisaea in Antarctica and impacts of climate change [Dataset]. Dryad.


Arctic Terns spend their breeding and non-breeding seasons in polar environments at opposite ends of the world. The sensitivity of polar regions to climate change makes it essential to understand the ecology of Arctic Terns but the remoteness of the Antarctic presents a considerable challenge. One solution is to use ‘biologgers’ to monitor remotely their behaviour and distribution in the Antarctic. Data from birds tagged with light-level global location sensors (geolocators) in 2015 and 2017 showed that a third of their annual cycle was spent amongst Antarctic sea ice. After reaching the East Antarctic in the austral spring, they gradually moved west, foraging in fragmented ice zones of the Antarctic coastline, leaving in the austral autumn for their return northward migration via the Atlantic. Changes in patterns of movement between phases of 24-h daylight and diel day/night conditions were likely linked to the annual moult, and stable isotope analyses suggest that krill (Euphausia species) was an important component of their diet. There were marked differences in movement behaviour between Arctic Terns tagged in 2015 compared to 2017 that may relate to unusual changes in sea-ice extent. The Arctic Tern may be unique amongst seabirds that utilise the Antarctic environment in summer in being able to move widely without nesting constraints, and may present a means of characterising the effects of climate change on species dependent for foraging on Antarctic sea-ice and krill.


Geolocator light-level data (maximum light-levels in 5 minute intervals referenced to GMT with the internal clock) represent 47 wintering profiles of geolocator deployments on Arctic Terns Sterna paradisaea from the Farne Islands (latitude 55.617; longitude -1.656), Northumberland in 2015 and 2017. The 23 geolocators from the 2015 cohort had been programmed to record either over the full light-level range (13 geolocators) or in ‘clipped’ mode where light-level data were sufficient to identify dawn and dusk thresholds accurately but temperature was not recorded. In the full light-range recording mode (used for all geolocators in 2017), minimum temperatures were recorded every 4 h. The deployment site was the calibration location. Four .csv files are provided for each geolocator deployment for geolocators programmed in full-light-level mode, and three .csv files for geolocator deployments where the geolocator was operating in clipped mode (no temperature data). For each .csv file, the first three characters (consisting of ‘G’ followed by a two-digit number) indicate the geolocator deployment, and subsequent characters indicate the file contents:

_calibration_period_lux_data.csv: raw date/time and light-level data for the calibration periods at the calibration site;

_lux_data_antarctic.csv: raw date/time and light-level data for the Antarctic between 16 October and 12 April next year;

_meta_data.csv: metadata for the raw light-level and temperature data, including geolocator type and operating characteristics;

_min-temp.csv: date/time and minimum temperature data (degrees Celsius) between 1 October and 15 April.

Raw light-level data can be interpreted using appropriate software (R version 3.3.1 and the R package FLightR were used in the paper to estimate position, applying use/stay criteria for land of 50 and 10 km, respectively). Twilight thresholds are necessary to estimate day/night length and the time of midnight and noon for the determination of latitude and longitude, respectively, but this is not possible in 24-h daylight when light levels are not reduced below the twilight thresholds. For birds in 24-h daylight, longitude can be estimated from light-level data by curve-fitting to estimate the time of midnight, represented by minimum daily light levels. For light-level data files, the columns are date/time and light levels (units). The files for one deployment, G77, differ in that in addition to raw date/time and light-level data, additional columns (twilight, interp and excluded) indicate dawn/dusk threshold assignments output from the ‘preprocessLight’ function of the R package BAStag program used for initial data processing.


Seabird Group

Natural History Society of Northumbria

Migrate Technology Ltd

BBC Springwatch