Breeding Experience Might Be a Major Determinant of Breeding Probability in Long-Lived Species: The Case of the Greater Flamingo

The probability of breeding is known to increase with age early in life in many long-lived species. This increase may be due to experience accumulated through past breeding attempts. Recent methodological advances allowing accounting for unobserved breeding episodes, we analyzed the encounter histories of 14716 greater flamingos over 25 years to get a detailed picture of the interactions of age and experience. Survival did not improve with experience, seemingly ruling out the selection hypothesis. Breeding probability varied within three levels of experience : no breeding experience, 1 experience, 2+ experiences. We fitted models with and without among-individual differences in breeding probabilities by including or not an additive individual random effect. Including the individual random effect improved the model fit less than including experience but the best model retained both. However, because modeling individual heterogeneity by means of an additive static individual random effect is currently criticized and may not be appropriate, we discuss the results with and without random effect. Without random effect, breeding probability of inexperienced birds was always times lower than that of same age experienced birds, and breeding probability increased more with one additional experience than with one additional year of age. With random effects, the advantage of experience was unequivocal only after age 9 while in young having experience was penalizing. Another pattern, that breeding probability of birds with experiences dropped after some age (8 without random effect; up to 11 with it), may point to differences in the timing of reproductive senescence or to the existence of a sensitive period for acquiring behavioral skills. Overall, the role of experience appears strong in this long-lived species. We argue that overlooking the role of experience may hamper detection of trade-offs and assessment of individual heterogeneity. However, manipulative experiments are desirable to confirm our finding.