Search

Population responses to environmental variation ultimately depend on within-individual and among-individual variation in labile phenotypic traits that affect fitness and resulting episodes of selection. Yet complex patterns of individual phenotypic variation arising within and between time periods, as well as associated variation in selection, have not been fully conceptualized or quantified. We highlight how structured patterns of phenotypic variation in dichotomous threshold traits can theoretically arise and experience varying forms of selection, shaping overall phenotypic dynamics. We then fit novel multistate models to 10 years of band-resighting data from European shags to quantify phenotypic variation and selection in a key threshold trait underlying spatioseasonal population dynamics: seasonal migration versus residence. First, we demonstrate substantial among-individual variation alongside substantial between-year individual repeatability in within-year phenotypic variation (“flexibility”), with weak sexual dimorphism. Second, we demonstrate that between-year individual variation in within-year phenotypes (“supraflexibility”) is structured and directional, consistent with the threshold trait model. Third, we demonstrate strong survival selection on within-year phenotypes—and hence on flexibility—that varies across years and sexes, including episodes of disruptive selection representing costs of flexibility. By quantitatively combining these results, we show how supraflexibility and survival selection on migratory flexibility jointly shape population-wide phenotypic dynamics of seasonal movement.

Dissecting joint micro-evolutionary and plastic responses to environmental perturbations fundamentally requires quantifying interacting components of genetic and environmental variation underlying expression of key traits. This ambition is particularly challenging for phenotypically discrete traits where multiscale decompositions are required to handle non-linear transformations of underlying genetic and environmental variation into phenotypic variation, especially when effects have to be estimated from incomplete field observations. We devised a novel joint multistate capture-recapture and quantitative genetic animal model, and fitted this model to full-annual-cycle resighting data from partially migratory European shags (Gulosus aristotelis) to estimate key components of genetic, environmental and phenotypic variance in the ecologically critical discrete trait of seasonal migration versus residence. We demonstrate non-trivial additive genetic variance in latent liability for migration, resulting in estimated micro-evolutionary responses following two episodes of strong survival selection. Yet, underlying additive genetic effects interacted with substantial permanent individual and temporary environmental effects to generate complex non-additive effects, causing large intrinsic gene-by-environment interaction variance in phenotypic expression. Our findings reveal how temporal dynamics of seasonal migration result from combinations of instantaneous micro-evolution and within-individual phenotypic inertia, and highlight how plastic phenotypic variation could expose cryptic genetic variation underlying discrete traits to complex forms of selection.

MOTIVATION: We undertook large citizen science surveys of bird distributions (atlases) in Britain and Ireland, aimed at quantifying breeding bird distributions on a 20‐year cycle and wintering bird distributions on a c. 30‐year cycle. We use these to generate spatially referenced information on apparent changes in bird distributions over c. 40 years. MAIN TYPE OF VARIABLE CONTAINED: Detection of breeding and wintering bird species in grid squares during five periods, and changes in detection between periods. The combined distribution dataset contains 1,410,938 records detailing detections of 465 bird species in 3,880 grid cells in different periods. The combined distribution change dataset contains 1,297,791 records describing stability, apparent colonization or apparent extinction of individual species in grid squares between pairs of atlases spanning up to c. 40 years. SPATIAL LOCATION AND GRAIN: Grid squares (10 km × 10 km) containing land throughout Britain, Ireland, the Isle of Man and the Channel Islands. The majority of data are at 10‐km resolution, but data for rare species are summarized at 20‐ or 50‐km resolution to protect sensitive locations. TIME PERIOD: The data represent summarized detection information derived from fieldwork during five periods: the breeding seasons 1968–1972, 1988–1991 and 2008–2011, and the winters 1981/1982–1983/1984 and 2007/2008–2010/2011. MAJOR TAXA STUDIED AND LEVEL OF MEASUREMENT: Birds; their distribution derived from citizen science surveys. SOFTWARE FORMAT: Data are supplied as comma‐separated text files.