Modeling the effects of age and breeding experience on the breeding probability of greater flamingos marked as chicks from 1977 to 1997 and resighted as breeders until 2001 in the Camargue, southern France.

Taking as a reference the fully age- and time-dependent survival and capture probabilities model [62], we calculated with program U-CARE [63] a variance inflation factor of 2.244 (section 1.1 of Appendix S3). In a first series of models, only the breeding probability part varied: first-year apparent survival in normal years was fixed to 0.763; first-year apparent survival for the cold spell year 1984 was estimated separately; adult apparent survival was estimated separately for the cold spell year and normal years (); capture probability was time-dependent (). For the variable breeding probability part, experience was a factor with 2 levels (): no experience, some previous experience; 3 levels (): no experience, 1 previous experience, previous experiences; or 4 levels (): no experience, 1 previous experience, 2 previous experiences, previous experiences; age () was a categorical variable. The notation ‘.’ means factorial model with main effects and interaction term. Our main model was with 3 levels of experience interacting fully with age. Two models had an additive individual random effect (). An additional model , , differed from the main model by an additive effect of experience on survival. is the Quasi Akaike Information Criterion relative to the main model; k is the model rank.