Skip to main content
Dryad logo

Sympatrically-breeding congeneric seabirds (Stercorarius spp.) from Arctic Canada migrate to four oceans


Harrison, Autumn-Lynn; Woodard, Paul; Mallory, Mark; Rausch, Jennie (2022), Sympatrically-breeding congeneric seabirds (Stercorarius spp.) from Arctic Canada migrate to four oceans, Dryad, Dataset,


Polar systems of avian migration remain unpredictable. For seabirds nesting in the Nearctic, it is often difficult to predict which of the world’s oceans birds will migrate to after breeding. Here we report on three related seabird species that migrated across four oceans following sympatric breeding at a central Canadian high Arctic nesting location. Using telemetry we tracked pomarine jaeger (Stercorarius pomarinus, n=1) to the Arctic Ocean to the western Pacific Ocean; parasitic jaeger (S. parasiticus, n=4) to the western Atlantic Ocean, and long-tailed jaeger (S. longicaudus, n=2) to the eastern Atlantic Ocean and western Indian Ocean. We also report on extensive nomadic movements over ocean during the post-breeding period (19,002 km) and over land and ocean during the pre-breeding period (5,578 km) by pomarine jaeger, an irruptive species whose full migrations and nomadic behavior have been a mystery. While the small sample sizes in our study limit the ability to make generalizable inferences, our results provide a key input to the knowledge of jaeger migrations. Understanding the routes and migratory divides of birds nesting in the Arctic region has implications for understanding both the glacial refugia of the past and the Anthropocene-driven changes in the future.


We captured adult jaegers during incubation (late June to early July) 2018 and 2019 at Nanuit Itillinga (Polar Bear Pass) National Wildlife Area, Bathurst Island, Nunavut, Canada (NINWA, 75° 43' N, 98° 24' W). We used 5 g (LTJA, n=2) and 9.5 g (PAJA, n=2 and POJA, n=1) Argos solar-powered satellite tags (Microwave Telemetry Inc., deployed 2018-2019) to track seabird movements. Satellite tags were attached using a leg-loop harness[22] made of 4.7625mm wide tubular Teflon Ribbon (Bally Ribbon Mills) secured with copper crimps. The total tag and attachment weight comprised 0.4-2.1% of the body mass of known-weight individuals (Table 1). We assessed wing and leg mobility prior to release and watched birds until they flew out of sight.

Data previously collected from two PAJA breeding on nearby Nasaruvaalik Island, Nunavut, Canada (58 km from NINWA, 75° 47' N, 96° 17' W) were also contributed to this study. These birds were tracked using archival light-level geolocators (GLS tags) attached with plastic cable ties to darvic leg-bands (Lotek Inc. LAT2900, 2.1g). Tags were deployed in July, 2010 (n=1) and June, 2011 (n=1) and recovered the following year by recapturing the birds. Tags also recorded sea surface temperature (SST) when immersed for more than 120 seconds and stored the minimum daily value.

Tag programming and processing

Satellite tags were duty-cycled to maximize solar charging (10 hours on, 48 hours off). Given sampling irregularity and the telemetry error of position estimates (see supplemental material for details), we used a model to estimate most probable paths. We applied the continuous-time random walk model of Jonsen et al.[23] using the foieGras package in R and estimated movement paths at 24-hour intervals to standardize sampling across birds tracked with different technologies (the maximum resolution of GLS tags was one position per day).

Light-level data from geolocator tags were initially processed using the manufacturer’s built-in template fit algorithm to estimate locations[24]. However, this method was shown to be biased south in winter and north in summer when applied to an Arctic seabird[25]. We therefore applied a sea surface temperature (SST) correction to further refine position estimates by comparing tag-collected SST measurements with remotely sensed SST data available for the same dates. We applied an unscented Kalman filter, a state-space model that incorporates measurement error estimation and the smoothing of the SST field directly in a single model to estimate the most probable track[26]. We formulated the model with a “solstice” error structure to account for highly erroneous positions near the equinoxes when light level is similar across the globe (defined by the model as September 16-October 2, March 10-March 27) and during which time positions were not estimated. Models were fit using the ukfsst package in R.

Usage Notes

The raw datasets in this study are available publicly under a creative commons license as a part of the Arctic Animal Movement Archive[52] on (Study Numbers: 973570814, 630339095, 300812056).