Differential colour-ring loss among Dalmatian Pelican (Pelecanus crispus) colonies and its consequence on survival estimates

The analysis of bird ringing data often comes with some potential sources of error and bias, as ring wear and/or loss could affect mark-recapture analyses and produce erroneous estimates of survival. Furthermore, ring wear and loss rates may differ between and within species based on the habitat they use or the species' life-history traits and behaviour as well as the type of the ring. In this study we use resighting data from a long-term double marking experiment to directly estimate the rate of colour-ring loss among different Dalmatian Pelican colonies over time, evaluate any possible factors that could contribute to differential ring loss and assess how it may bias the results of mark-resighting analyses. Based on 14,849 resightings from 1275 individuals and using multi-state continuous-time hidden Markov models (HMMs) we showed that probability of ring loss was markedly different among colonies, ranging from 0.10 to 0.42 within the first year of marking, whereas the cumulative probability of losing a ring after ten years ranged 0.64 to 0.99. These rates are among the highest estimated when compared to previous studies in waterbirds. Our approach assessing the intra-specific variance in ring loss provided several factors potentially involved, such as the use of glue and the fledgling age accuracy and we could further hypothesise the effect of environmental factors. Finally, our results showed that ring loss can be a significant challenge for the assessment of the species' population dynamics using mark-recapture methods as survival was consistently underestimated when not accounting for ring loss and varied significantly among different colonies.