Early‐life stasis in partial seasonal migration is underpinned by among‐cohort variation in migratory plasticity and selective disappearance

•1. Life‐history traits expressed in early life can exhibit considerable among‐cohort variation, which could substantially affect population age‐structure and dynamics if initial variation persists into later life‐stages. Yet, initial among‐cohort variation could be reinforced, rapidly dissipated, or else completely reshaped by dynamic combinations of age‐specific phenotypic plasticity and selective disappearance acting within and among cohorts. However, such effects have not been comprehensively quantified for any trait, precluding full prediction of the form and implications of phenotypic dynamics, and emerging age‐specific life‐history variation, in varying environments. •2. We provide a framework for conceptualising phenotypic change resulting from joint and interacting effects of cohort‐specific and age‐specific plasticity and selective disappearance. We implement this framework by quantifying overall early‐life age‐specific phenotypic change (or stasis), and dissecting underlying dynamics of plasticity and selection, for the ecologically critical life‐history trait of seasonal migration versus residence. We achieve this by fitting multi‐state models to extensive multi‐year ring‐resighting data from 9358 colour‐ringed European shags ( Gulosus aristotelis ) from 11 cohorts in a partially migratory population. •3. The overall cross‐cohort mean proportion of migrants versus residents remained approximately constant across the four winters following fledging, implying no overall change in the degree of seasonal migration with age. This stasis was underlain by consistently high cross‐year individual phenotypic repeatability, and by average plasticity towards residence that was counter‐acted by average selective disappearance of sub‐adult residents. •4. However, these cross‐cohort means obscured substantial among‐cohort variation in the initial degree of partial migration, and in subsequent joint effects of plasticity and selective disappearance. Here, plasticity and selection were not systematically associated within or across cohorts or ages, but rather reinforced versus counter‐acted each other at different times, thereby reshaping the pattern of among‐cohort variation in partial migration across ages. •5. These results demonstrate that an absence of overall age‐specific change in a key life‐history trait, seasonal migration versus residence, obscures substantial underlying variation in both early‐life plasticity and selective disappearance, generating complex phenotypic dynamics within individual cohorts. Standard cross‐cohort analyses may therefore inadequately predict future spatio‐seasonal dynamics, since novel age‐specific life‐histories could readily emerge given changing environmental drivers of plasticity and selection.