| Summary: | Three different mark-recapture models were applied to resighting data for Brünnich's guillemot (Uria lomvia) chicks obtained from photo series. Performances of the various models as potential tools for estimating chick survival over the breeding season were assessed. Of particular interest was Pradel’s multievent model, estimating state transition probabilities while accounting for uncertainty of the state of the chicks when only the adults were observed. The performances of the Cormack-Jolly-Seber (CJS) model and Pollock’s robust design were also explored. Estimates produced by all models were compared to field observations of breeding success. All of the models tested had difficulties coping with the strong heterogeneity in capture probabilities among chicks on wide and densely populated ledges. These conditions caused underestimated survival probabilities. Strong site fidelity exhibited by adult Brünnich's guillemots despite chick loss was another potential cause of bias in the multievent model estimators. The CJS model and Pollock’s robust design produced realistic survival rate estimates when heterogeneity in capture probabilities was low or could be modeled. Although sample size was relatively small, this study provides a good basis for further investigations, as it identifies the main challenges facing investigators planning to estimate chick survival in cliff-nesting seabirds using resighting data from photo series.
|
|---|