Design issues relevant to capture–recapture/resighting studies include timing
of sample periods and spatial coverage of sampled areas. Sample periods
themselves are the periods during which birds are captured, recaptured, and
re-observed, and may range from durations of 1 day (e.g. for estimation of weekly
postfledging survival rates on local areas, Krementz et al. 1989) to 2–3 months
(e.g. for estimation of annual survival rates). Sample periods typically should be
short relative to the intervals that separate them and should occur during seasons
when birds are relatively stationary and not migrating. Specifically, the sample
periods should be sufficiently short that they include little mortality, but suffi-
ciently long that the entire study area can be searched for birds or subjected
to capture efforts. The spatial sampling of the study area should be such that
all birds on the area should have similar, a prioriprobabilities of being caught
or observed each sample period. If the entire area cannot be covered in each
sampling period, then the sections to be covered can be randomly selected at
each sample period. Investigators should ensure that certain portions of the
study area are not consistently avoided or poorly sampled.
Temporary emigration from the study area can cause problems with the
estimation of survival rate. Kendall et al. (1997) found that these problems can
be largely remedied by collecting data under Pollock’s (1982) robust design.
Here sampling for each primary period of interest (e.g. the sample periods
discussed thus far) consists of at least two distinct secondary capture sessions that
encompass the entire study area and are closely spaced in time. For example, it
might take the investigator 4 days to sample all portions of a local study area,
so the investigator might conduct such 4-day sampling during each of three
consecutive weeks in a breeding season. Each 4-day sample period would be con-
sidered a secondary sampling period, whereas the three sets of such periods
combined would constitute the primary sample period for the year. Study design
for survival rate estimation with open-population and robust design models is
discussed in more detail by Williams et al. (2002), and figures for use in selecting
needed sampling intensities are provided by Pollock et al. (1990).
5.3.3Band recovery
Band recovery studies generally involve the application of metal legbands to
birds at one or more study locations where recaptures of previously banded birds
are either infrequent or ignored. Estimation is instead based on bands recovered
by members of the public, often throughout the range of the bird. Typical cases
in which recoveries far outnumber recaptures include banding of waterfowl on
breeding grounds or at molting sites, with subsequent recoveries coming from
hunters who report bands from birds they shoot on migration and wintering
Survival rates| 127