Résumé
Migrating animals often divide their journey into alternating phases of migration bouts and stopping over. For investigating many questions of migration ecology it is crucial (1) to estimate the duration of stopover phases, and (2) to test whether animals of different groups differ in their stopover behavior. Using recent advances in capture-recapture statistics, we show how total stopover duration can be estimated from capture-recapture data. The probabilities of immigration are estimated and modeled by recruitment analysis and are converted into the time the animals spent at the stopover place before capture; the probabilities of emigration are estimated and modeled by survival analysis and are converted into the time the animals spent at the stopover place after capture. The sum of the two parts is the total stopover duration. Tests for differences between groups can be addressed by an appropriate model selection procedure. Two examples of migrating passerine birds at a stopover site in Switzerland illustrate this method. Mean total stopover duration was 12.3 d for Reed Warblers and 7.1 d for Reed Buntings. This was considerably higher than values obtained by the minimum stopover duration estimation (6.0 and 4.4 d, respectively). Because of the fundamental weaknesses of the minimum stopover duration estimation, which has been widely used in migration ecology, many findings obtained by this method need to be reconsidered.