possible to mark large numbers of birds, the following type of question is likely
to arise: should I mark 30 animals with radios or 300 animals with conventional
tags? Informed answers to such questions will require pilot data or guesses about
capture-resighting probabilities and development of simulation- or approximation-
based sample size figures such as those of Pollock et al. (1990) that can be used to
compare estimator precision under different scenarios. However, estimator pre-
cision is not the only quantity of relevance to such study design decisions, as
potential for estimator bias (e.g. via radio effects on survival, or dependence
of fate and censoring) will also be relevant. As noted previously, mechanical
problems and difficulty in detecting signals in some habitats and sampling
situations can also lead to radio-telemetry detection probabilities 1.
When field methods do not permit all marked birds to be detected, then study
designs should seek to minimize variation in detection probabilities among
marked birds and to identify and record important sources of variation (e.g. bird
location, bird sex) that still exist. The use of Pollock’s (1982) robust design offers
several advantages for capture–recapture studies, including the ability to account
for movement to areas outside the study area(s). Band recovery models tend to be
most useful when large numbers of birds (e.g. 100s to 1000s) can be banded and
when recovery rates (probability that a banded bird alive at the time of banding
dies, is found and has its band reported) are relatively high (e.g. 0.04). Such
high recovery rates typically occur only for hunted species. Recapture and
resighting probabilities in many intensive capture–recapture studies of birds are
fairly high (e.g. 0.2 and sometimes 0.5) and permit precise estimation
of apparent survival. Once again, however, precision is not the only relevant
quantity. The complement of apparent survival from capture–recapture studies
includes both death and permanent emigration. The permanent emigration
component may be small for adult birds but is frequently very large for young
birds. The complement of survival estimates resulting from most band recovery
studies includes only death, so there is an advantage to the use of such estimators
when recovery rates are sufficiently large.
In general, studies directed at questions about survival or movement should
be designed in a manner that exploits available field methods and their respect-
ive analytic and modeling approaches. Rather than focusing exclusively on the
selection of single data types and associated designs, it is becoming increas-
ingly clear that hybrid designs offer many advantages. Combining sources of
information, such as capture/resighting, radio-telemetry, recoveries, or sightings
between formal sampling periods, can be used to address questions about bird sur-
vival and movement, sometimes permitting estimation of otherwise inaccessible
parameters, as well as increasing the precision of estimates (e.g. see Barker 1997;
Summary and general recommendations| 135