Wildfowl 63 - page 82

76 Survival and dispersal in a new Barnacle Goose population
©Wildfowl & Wetlands Trust Wildfowl (2013) 63: 72–89
when the time-span was slightly shorter.
This shorter (c. 6–9 months) survival
duration estimated for first-year birds was
unavoidable, since grouping the re-sightings
differently (e.g . from July–June, instead of
from January–December) would lead to the
median month of the first observation
being in December or January, and an even
shorter time-span during the first year. The
use of a two-age class structure introduced
for both groups accounted for the
discrepancy in time span between the first
and second age-class.
Data from birds ringed at the two
localities in the Delta area were analysed
separately from data on birds ringed in
Fryslân in 2009. Because dead recoveries of
(mainly shot) birds were available for geese
ringed in the Delta area, in addition to the
live observations, the Burnham model was
used as this estimates survival using both re-
sightings and recovery data simultaneously.
Data from Fryslân were analysed with a
standard Cormack-Jolly-Seber (CJS) model
using re-sightings of live birds only. First,
the most parsimonious model was found for
the re-sightings data, incorporating resighting
rate (p) and, in the delta area, the reporting rate
(r) and the fidelity parameter (F), prior to
estimating survival rates. Additional models
based on the initial model then tested
whether survival was time-, group-, or age-
dependent, with group reflecting the bird’s
age on ringing. In order to test whether
survival during the first year differed from
survival rates thereafter, survival in age class
2 (i.e . birds ringed as juveniles) was set as
being equal to the survival of birds ringed as
adults in both age classes, to make optimal
use of all available data. To test for
overdispersion, cˆ was calculated to be 1.247
based on 100 simulations in MARK. AIC
and Deviance values therefore were
subsequently adjusted with this value.
Dispersal
The dispersal distance from the site of
ringing to a breeding area in one of
subsequent years was calculated using all
20,692 observations. For each observation,
the distance between the site of ringing (the
place of birth or the breeding / moulting
site) and the site of breeding was calculated
as the great circle distance between the two
sited, using the following formula:
Distance = 6372.8 × arccos (sin latr × sin
latw + cos latr × cos latw × cos
Δlong)
where:
(latr, longr) = latitude and longitude of
ringing site in radians (π × lat /
180, π × long / 180)
(latw, longw) = latitude and longitude of
breeding site in radians
(π × lat / 180, π × long / 180)
Δlong = difference between longitude
of ringing and breeding site,
and
6372.8 = radius of the earth in km.
It was assumed that all sightings made of
Barnacle Geese in the Netherlands between
1 April–15 August were in (potential)
breeding habitat, and that birds ringed as
adults or birds ringed as juveniles but at least
two years old on the observation date,
attempted or at least intended to breed
there. The probability of Dutch-breeding