Stratification based on reproductive state reveals contrasting patterns of age‐related variation in demographic parameters in the kittiwake

Abstract

Heterogeneity in individual quality can be a major obstacle when interpreting age‐specific variation in life‐history traits. Heterogeneity is likely to lead to within‐generation selection, and patterns observed at the population level may result from the combination of hidden patterns specific to subpopulations. Population‐level patterns are not relevant to hypotheses concerning the evolution of age‐specific reproductive strategies if they differ from patterns at the individual level. We addressed the influence of age and a variable used as a surrogate of quality (yearly reproductive state) on survival and breeding probability in the kittiwake. We found evidence of an effect of age and quality on both demographic parameters. Patterns observed in breeders are consistent with the selection hypothesis, which predicts age‐related increases in survival and traits positively correlated with survival. Our results also reveal unexpected age effects specific to subgroups: the influence of age on survival and future breeding probability is not the same in nonbreeders and breeders. These patterns are observed in higher‐quality breeding habitats, where the influence of extrinsic factors on breeding state is the weakest. Moreover, there is slight evidence of an influence of sex on breeding probability (not on survival), but the same overall pattern is observed in both sexes. Our results support the hypothesis that age‐related variation in demographic parameters observed at the population level is partly shaped by heterogeneity among individuals. They also suggest processes specific to subpopulations. Recent theoretical developments lay emphasis on integration of sources of heterogeneity in optimization models to account for apparently “sub‐optimal” empirical patterns. Incorporation of sources of heterogeneity is also the key to investigation of age‐related reproductive strategies in heterogeneous populations. Thwarting “heterogeneity's ruses” has become a major challenge: for detecting and understanding natural processes, and a constructive confrontation between empirical and theoretical studies.

Now you see him, now you don't: experience, not age, is related to reproduction in kittiwakes

“…We see several explanations for this apparently contradictory result. An age effect can be masked by individual heterogeneity [11], which was not accounted for here, or may be detectable only when accounting for other covariates [23]. Alternatively, while many experienced individuals are reobserved at oldest ages, none of the inexperienced individuals reached old ages.…”

“…Again, experience was not taken into account, and the analysis required selection of a specific subset of individuals confined in a given spatial unit. In addition, it was previously suggested that the approach used to define clusters of individuals breeding in the same spatial units (cliffs) was not satisfactory [23]. Indeed, the number of pairs in such units ranged from 10 to 300; unless there was no heterogeneity in breeding activity and success in the unit (e.g.…”

“…A previous study accounted for the covariance between individuals breeding on the same cliff to address the relationship between previous breeding state and future breeding probability or between age and breeding probability [23]. Again, experience was not taken into account, and the analysis required selection of a specific subset of individuals confined in a given spatial unit.…”

“…members of a pair could be treated as clusters in mixed models; [23,43]). An earlier study of this population did not provide evidence that results concerning the relationship between previous breeding state and subsequent breeding probability on the one hand, or between age and breeding probability on the other hand, were sensitive to incorporation of a covariance between members of a pair [23]. However, experience had not been considered, and the sample used concerned only members of double-marked pairs.…”

“…Note, however, that this unusually high detection probability is not a prerequisite to implement the statistical approach; we propose to assess breeding experience (see §2b). Non-breeders do not always establish in a specific place (not confined to a colony or cliff [23,24]), or at least cannot be assigned to a specific location because the resightings are almost equally distributed among different places. However, it is sometimes possible to assign them to a specific site or colony (non-breeders confined to a cliff) if they consistently attend a place, exhibit specific territorial behaviour or get involved in the preliminary steps of the breeding process with a mate (e.g.…”