Among-year variation in the repeatability, within- and between-individual, and phenotypic correlations of behaviors in a natural population

When mean behaviors correlate among individuals, they form behavioral syndromes. One way to understand the evolution of such a group-level phenomenon is to compare horizontally patterns of correlations among populations (or species) or follow longitudinally the same population over years in the light of parallel differences in the environment. We applied the longitudinal approach to 8-year field data and analyzed phenotypic correlations, and their within- and between-individual components, among three behaviors (novelty avoidance, aggression, and risk-taking) in male collared flycatchers, Ficedula albicollis, in a meta-analytic framework. The phenotypic correlation between novelty avoidance and aggression varied heterogeneously (it was positive in some years, while it was negative in other years), while the other pair-wise correlations were consistently positive over the study period. We investigated four potential socio-ecological factors, and found evidence that the among-year alterations in the demographic structure of the population (density, age composition) can be responsible for the heterogeneous effect sizes. Comparing within- and between-individual correlations across pairs of traits, we found that the correlation between aggression and risk-taking at the among-individual level was the strongest suggesting that this relationship has the highest potential to form a behavioral syndrome. Within-year repeatabilities varied among traits, but were systematically higher than between-year repeatabilities. Our study highlights on an empirical basis that there can be several biological and statistical reasons behind detecting a phenotypic correlation in a study, but only few of these imply that fixed behavioral syndromes are maintained in a natural population. In fact, some correlations seem to be shaped by environmental fluctuations.