WINTER MOVEMENT ANALYSIS
METHODS
The first step to analyse the winter movements is to define the winter period for each individual. Since King Eiders winter over a broad latitudinal range winter does not start at the same time for all birds. We will use the time between migratory ‘seasons’ as a definition for winter period, that is, the time between the end of fall migration and the onset of spring migration. This period ranges from 56 to 280 days between different individuals. Some birds do not have an actual fall migration and thus arrive on their wintering grounds after their molt migration in late July or August.
We only consider movements exceeding 50 km in this analysis, as shorter distance movements are too common and may be motivated by local disturbances or patchy food resources that we cannot track.
For every movement we recorded whether other birds with satellite transmitters were in the vicinity of the departure location during the time the focal bird departed from a site. This was the case in over 70% of our movements. We then classified whether other birds in the vicinity departed at the same time, or stayed in the area to test whether winter movements occurred simultaneously.
Sea Ice data for the Bering Sea is available from the National Ice Center, and we downloaded sea ice maps for every day marking the beginning and end of a King Eider movement. We then overlaid the ducks position with the sea ice map in GIS and extracted the information of the sea ice concentration at the point of departure before and after the duck had left this site. We then calculated the change in sea ice in the same way for all duck locations from which the ducks did not depart .
To examine which factors are correlated with a King Eider's decision to move or to stay we constructed a Random Forest model with several predictor variables (latitude, day length, sea ice concentration, benthic biomass, sex, year, individual bird).