Skip to main content
Dryad logo

Data from: Resource predictability drives inter-annual variation in migratory behaviour in a long-lived bird


Baert, Jan et al. (2021), Data from: Resource predictability drives inter-annual variation in migratory behaviour in a long-lived bird, Dryad, Dataset,


There is a growing awareness that experience may play a major role in migratory decisions, especially in long-lived species. However, empirical support remains to date scarce. Here, we use multi-year GPS-tracking data on 28 adult Lesser Black-backed Gulls (Larus fuscus), a long-lived species for which migratory strategies typically consist of a series of long stopovers, to assess how experience affects interannual variation in stopover selection. We expect that food source reliability should play a pivotal role, as it both reduces the uncertainty on food availability across years, and enables for more efficient foraging during stopovers by reducing searching efforts. We found that during stopovers gulls indeed developed high fidelity to particular foraging locations, which strongly reduced the daily distance travelled for foraging. When revisiting stopovers in consecutive years, birds used over 80% of foraging locations from the previous year. Although the average fidelity to stopovers across years was a high as 85%, stopovers where birds showed high foraging site fidelity were up to 60% more likely to be revisited compared to stopover with low foraging site fidelity. Accordingly, birds using more stopovers with reliable foraging opportunities showed significantly less interannual variation in their stopover use than birds using stopovers with less reliable foraging opportunities. Our results thus highlight the need to further deepen our understanding of the role of cognitive processes in individual variation in migratory behaviour.


Calculated metrics for foraging behaviour and interannual stopover fidelity based on GPS data collected on 28 adult Lesser Black-backed Gulls. Raw GPS data for these birds are publicly available at Zenodo ( First, data were subsampled to a 30’ resolution. Next, data were split into stopovers and migratory bouts using density-based spatial clustering of applications with noise (DBSCAN) with a 0.1 degrees neighbourhood radius. As birds typically migrate at 30–40km per hour, this ensures that points do not cluster during the unidirectional movement of migration but only when birds are performing repetitive movements during stopovers. Subsequently, we identified putative foraging locations for each day during stopovers based on a ground speed below 5m s-1. Similar to foraging, roosting also involves very low ground speeds. We identified these night roosts, on which birds where not assumed to forage, as the last location for each day where birds were inactive for at least 3 hours uninterrupted. We considered birds to be inactive if the instantaneous ground speed measured by the accelerometer of the trackers did not exceed 0.5 m s-1 and birds stayed within a 1 km radius for at least 3 hours.

For each day during a stopover, we calculated the total distance covered, the fidelity to foraging sites and, for stopovers that have been revisited in the previous year, the reuse of foraging sites from the previous year. The total distance covered was calculated as the summed haversine distance between all GPS positions for a given day, and corrected for the coverage (i.e. the ratio between the number of GPS positions and the maximum possible number of positions, which is 48). The fidelity to foraging sites during a stopover was calculated as the proportion of GPS positions where the bird was assumed to be foraging, which lay within 500 m of a GPS position where the bird had been foraging within the previous 7 days. Similarly, for revisited stopovers, the reuse of foraging sites from the previous year was calculated as the proportion of GPS positions which were considered to be foraging, which lay within 500 m of a GPS position where birds had been foraging in the previous year. Stopovers were considered to be revisited when GPS positions of stopovers overlapped in consecutive years. Metrics were not calculated for days with less than 12 hours of data.

Usage Notes

File description

bird: bird nickname

ringnumber: metal ring inscription

cycle: migration cycle, referring to the year in which the bird leaft the colony

year: the year from 'migration cycle'

colony: colony from which the bird originates

date: date for which the measures are calculated

breeding: indicating weather a birds is breeding or has left the breeding areas. Note that a bird is considered breeding as long as it regularly visits the colony

max.mig.dist: maximum distance a birds migrates from the colony during its migration cycle (km)

dist.col: distance of a stopover from the colony (km) days since arrival on a stopover

stop.dur: total stopover duration (day)

dist.cov: daily distance covered (km)

revist: boolean indicating if the foraging sites used are a revisit from previous days

revisit.prev: boolean indicating if the foraging sites used are a revisit from the previous year

stop.rev: boolean indicating a a stopover has been visited in the previous year

coverage: hours of data coverage on which metrics are based (h)


FWO, Award: 12R7619N

Fonds Wetenschappelijk Onderzoek, Award: G0E1614N

Universiteit Gent, Award: 01M00221