Population dynamic and eco-evolutionary responses to environmental variation and change fundamentally depend on combinations of within- and among-cohort variation in phenotypic expression of key life-history traits, and on corresponding variation in selection on those traits. Specifically, in partially migratory populations, spatio-seasonal dynamics depend on the degree of adaptive phenotypic expression of seasonal migration versus residence, where more individuals migrate when selection favours migration. Opportunity for adaptive (or, conversely, maladaptive) expression could be particularly substantial in early life, through initial development of migration versus residence. However, within- and among-cohort dynamics of early-life migration, and of associated survival selection, have not been quantified in any system, preventing any inference on adaptive early-life expression. Such analyses have been precluded because data on seasonal movements and survival of sufficient young individuals, across multiple cohorts, have not been collected. We undertook extensive year-round field resightings of 9,359 colour-ringed juvenile European Shags (Gulosus aristotelis) from 11 successive cohorts in a partially-migratory population. We fitted advanced Bayesian multi-state capture-mark-recapture models to quantify early-life variation in migration versus residence and associated survival across short temporal occasions through each cohort's first year from fledging, thereby quantifying the degree of adaptive phenotypic expression of migration within and across years. All cohorts were highly partially migratory, but the degree and timing of migration varied considerably within and among cohorts. Episodes of strong survival selection on migration versus residence occurred both on short timeframes within years, and cumulatively across whole years, generating instances of instantaneous and cumulative net selection that would be obscured at coarser temporal resolutions. Further, the magnitude and direction of selection varied among years, generating strong fluctuating survival selection on early-life migration across cohorts, as rarely evidenced in nature. Yet, the degree of migration did not strongly covary with the direction of selection, indicating limited early-life adaptive phenotypic expression. These results reveal how dynamic early-life expression and selection on a key life-history trait, seasonal migration, can emerge across seasonal, annual, and multi-year timeframes, yet be substantially decoupled. This restricts the potential for adaptive phenotypic, micro-evolutionary, and population dynamic responses to changing seasonal environments.

Funding provided by: The Research Council of NorwayROR ID: https://ror.org/00epmv149Award Number: 223257 Funding provided by: Natural Environment Research CouncilROR ID: https://ror.org/02b5d8509Award Number: NE/M005186/1 Funding provided by: Natural Environment Research CouncilROR ID: https://ror.org/02b5d8509Award Number: NE/R000859/1 Funding provided by: Natural Environment Research CouncilROR ID: https://ror.org/02b5d8509Award Number: NE/R016429/1 Funding provided by: The Research Council of NorwayROR ID: https://ror.org/00epmv149Award Number: 313570

The data come from a long-term study of European shags (Gulosus aristotelis) breeding the Isle of May National Nature Reserve, Scotland (56°11'5.40"N, 2°33'16.19"W). The data comprise capture-recapture history data derived from field-ring-resighting data from 11 cohorts of first-year juveniles between 2010 and 2020. Codes for bespoke multi-state capture-mark-recapture models are provided.