Recapture processes and biological inference in monitoring burrow-nesting seabirds

Capture-mark-recapture methods are used widely for monitoring and diagnosis of bird populations as they permit robust estimates of population abundance and demographic parameters (e.g. survival) to be obtained from incomplete records of individual life histories. The statistical analysis of these data relies on the important assumption that individuals of the same local populations (i.e. colony) have the same parameters (the homogeneity assumption). We used data from six medium- to long-term monitoring schemes of local Mediterranean populations of the European Storm Petrel Hydrobates pelagicus to empirically show that the level of individual heterogeneity and the consequent bias in the parameter of interest depend on the recapture methodology, which has important consequences for the experimental design. We found that the recapture probability varied over time and among methodologies. The study design had a strong influence on the proportion of transients caught (i.e. individuals not recaptured after marking); however, the survival probability estimate for resident birds was fairly similar across the studies. The differences found in survival seem to depend on the biological variability between sites (e.g. predation pressure), and not on the recapture methods.