Age-dependent trait variation: the relative contribution of within-individual change, selective appearance and...

Within populations, the expression of phenotypic traits typically varies with age. Such age-dependent trait variation can be caused by within-individual change (improvement, senescence, terminal effects) and/or selective (dis)appearance of certain phenotypes among older age classes. In this study, we applied two methods (decomposition and mixed modelling) to attribute age-dependent variation in seven phenological and reproductive traits to within-individual change and selective (dis)appearance, in a long-lived seabird, the common tern (Sterna hirundo). At the population level, all traits, except the probability to breed, improved with age (i.e. phenology advanced and reproductive output increased). Both methods identified within-individual change as the main responsible process, and, within individuals, performance improved until age 6-13, before levelling off. In contrast, within individuals, breeding probability decreased to age 10, then levelled off. Effects of selective appearance and disappearance were small, but showed that longer-lived individuals had a higher breeding probability and bred earlier and that younger recruits performed better throughout life than older recruits in terms of both phenology and reproductive performance. In the year prior to death, individuals advanced reproduction, suggesting terminal investment. The decomposition method attributed more age-dependent trait variation to selective disappearance than the mixed-modelling method: 14-36% versus 0-8%, respectively, which we identify to be due to covariance between rates of within-individual change and selective (dis)appearance leading to biased results from the decomposition method. We conclude that the decomposition method is ideal for visualizing processes underlying population change in performance from one age class to the next, but that a mixed-modelling method is required to investigate the significance and relative contribution of age effects. Considerable variation in the contribution of the different age processes between the seven phenotypic traits studied, as well as notable differences between species in patterns of age-dependent trait expression, calls for better predictions regarding optimal allocation strategies with age.