permanent emigration, it is not clear how to model a bird with which radio
contact is simply lost during the study. In some cases, inferences about this prob-
lem are possible when the study also includes birds marked with standard tags,
permitting estimation of survival with capture–recapture/resighting data as well
as with telemetry (see Bennetts et al. 1999). As in any effort to estimate para-
meters of natural animal populations, the primary recommendation is to tailor
the estimation model to the realities of the sampling process as much as possible.
5.3.2Capture–recapture/resighting
Capture–recapture and resighting studies are often carried out in single, local
study areas, and data collection is usually by investigators and not by the general
public. The duration of the sampling period and the length of the interval
between sampling occasions can vary substantially and depend on study object-
ives. Sampling 1 day each week has been used to estimate postfledging sur-
vival (e.g. Krementz et al. 1989), whereas most studies involve longer sampling
occasions and intervals. Many studies of bird populations are carried out during
the breeding season, producing multi-year data sets with sampling periods of
6–8 weeks each year. If standard metal legbands are used to mark birds, then sampl-
ing typically involves recapturing birds each year (e.g. with mist nets, nest traps,
rocket nets, etc.). If colored legbands, neck collars (e.g. for geese and swans), nasal
discs (e.g. for ducks) or patagial tags are applied, then resampling may involve
observations of individual birds with spotting scopes or binoculars. In the remain-
der of this section, we will refer to capture–recapture models with the under-
standing that this is a general descriptor that pertains also to resighting studies.
Data resulting from a capture–recapture study differ from radio-telemetry
data in that deaths are not typically observed in the former type of study. Instead,
it is possible to estimate a parameter frequently termed “apparent survival” or
“local survival” to emphasize the fact that its complement includes both death
and permanent emigration from the study area. The data are typically summar-
ized as individual capture histories, which are simply rows of 1’s and 0’s indicat-
ing whether each bird is (1) or is not (0) captured at each sampling period of the
study. For example, a 5-year study of a breeding population might yield one or
more birds with the following capture history: 0 1 0 1 0. The five entries in the
row represent the 5 years of the study. A bird with this history was first captured
during sampling in year 2 of the study. It was marked and released following
capture, was not detected in year 3, was recaptured or resighted in year 4 and
not detected in the final year of the study, 5.
As in the modeling of radio-telemetry data, the key to estimation of survival
parameters from capture–recapture data is to develop a reasonable model of the
124 |Estimating survival and movement