Population and evolutionary dynamics in spatially structured seasonally varying environments

Abstract

Increasingly imperative objectives in ecology are to understand and forecast population dynamic and evolutionary responses to seasonal environmental variation and change. Such population and evolutionary dynamics result from immediate and lagged responses of all key life-history traits, and resulting demographic rates that affect population growth rate, to seasonal environmental conditions and population density. However, existing population dynamic and eco-evolutionary theory and models have not yet fully encompassed within-individual and among-individual variation, covariation, structure and heterogeneity, and ongoing evolution, in a critical life-history trait that allows individuals to respond to seasonal environmental conditions: seasonal migration. Meanwhile, empirical studies aided by new animal-tracking technologies are increasingly demonstrating substantial within-population variation in the occurrence and form of migration versus year-round residence, generating diverse forms of 'partial migration' spanning diverse species, habitats and spatial scales. Such partially migratory systems form a continuum between the extreme scenarios of full migration and full year-round residence, and are commonplace in nature. Here, we first review basic scenarios of partial migration and associated models designed to identify conditions that facilitate the maintenance of migratory polymorphism. We highlight that such models have been fundamental to the development of partial migration theory, but are spatially and demographically simplistic compared to the rich bodies of population dynamic theory and models that consider spatially structured populations with dispersal but no migration, or consider populations experiencing strong seasonality and full obligate migration. Second, to provide an overarching conceptual framework for spatio-temporal population dynamics, we define a 'partially migratory meta-population' system as a spatially structured set of locations that can be occupied by different sets of resident and migrant individuals in different seasons, and where locations that can support reproduction can also be linked by dispersal. We outline key forms of within-individual and among-individual variation and structure in migration that could arise within such systems and interact with variation in individual survival, reproduction and dispersal to create complex population dynamics and evolutionary responses across locations, seasons, years and generations. Third, we review approaches by which population dynamic and eco-evolutionary models could be developed to test hypotheses regarding the dynamics and persistence of partially migratory meta-populations given diverse forms of seasonal environmental variation and change, and to forecast system-specific dynamics. To demonstrate one such approach, we use an evolutionary individual-based model to illustrate that multiple forms of partial migration can readily co-exist in a simple spatially structured landscape. Finally, we summarise recent empirical studies that demonstrate key components of demographic structure in partial migration, and demonstrate diverse associations with reproduction and survival. We thereby identify key theoretical and empirical knowledge gaps that remain, and consider multiple complementary approaches by which these gaps can be filled in order to elucidate population dynamic and eco-evolutionary responses to spatio-temporal seasonal environmental variation and change.

Keywords: demographic structure; density-dependence; eco-evolutionary dynamics; life-history variation; meta-population; movement ecology; partial migration; plasticity; population viability; seasonal migrant; vital rate.

Figure 1

(A) Illustration that seasonal partial migration forms a continuum between the extreme scenarios of full obligate migration and full year‐round residence. (B–J) Examples of diverse partially migratory taxa. Migratory individuals can make long‐ or medium‐distance geographical migrations (B–E), or medium‐distance altitudinal migrations (F–H), or short‐distance migrations between adjacent but distinct habitat types (I, J). Geographical migrations occur in (B) wandering albatross (Diomedea exulans; Weimerskirch et al., 2015); (C) tiger shark (Galeocerdo cuvier; Papastamatiou et al., 2013); (D) European shag (Phalacrocorax aristotelis; Grist et al., 2014, 2017); (E) Skylark (Alauda arvensis; Hegemann et al., 2015), and numerous other birds [including blackbirds, Turdus merula (Fudickar et al., 2013; Zúñiga et al., 2017), and American kestrels, Falco sparverius (Anderson et al., 2015)]. Altitudinal migrations occur in (F) elk (Cervus elaphus; Hebblewhite & Merrill, 2011; Eggeman et al., 2016) and other ungulates (e.g. caribou Rangifer tarandus; McDevitt et al., 2009), and in (G) white‐ruffed manakin (Corapipo altera; Boyle et al., 2010, 2011) and (H) American dipper (Cinclus mexicanus; Gillis et al., 2008; Green et al., 2015) and numerous other birds (Boyle, 2017). Habitat‐related migrations occur in (I) red‐spotted newt (Notophthalmus viridescens, Grayson & Wilbur, 2009; Grayson et al., 2011) and (J) roach (Rutilus rutilus; Brodersen et al., 2008; Skov et al., 2013) and many other fish (e.g. Chapman et al., 2012; Vélez‐Espino et al., 2013), and also ungulates such as roe deer (Capreolus capreolus; Peters et al., 2017). Seasonal partial migration across diverse spatial scales also occurs in reptiles (e.g. Shaw & Levin, 2011; Yackulic et al., 2017). Partial migration can also occur on shorter timeframes, including diel migrations observed in fish and invertebrates (e.g. Chapman et al., 2011; Harrison et al., 2017). Photograph credits: (B) Henri Weimerskirch; (C) Yannis Papastamatiou; (D) Mark Newell; (E) Rob Voesten; (F) Celie Intering; (G) Alice Boyle; (H) Roberta Olenick; (I) Kristine Grayson; (J) Jakob Brodersen.

Figure 2

Illustration of four basic scenarios of partial migration considered as mutually exclusive alternatives. Resident (R) and migrant (M) individuals can (A) co‐exist in the same location in the breeding season but be spatially separated in the non‐breeding season (‘non‐breeding partial migration’, also known as ‘shared breeding partial migration’); or (B) co‐exist in the non‐breeding season but be spatially separated in the breeding season (‘breeding partial migration’, also known as ‘shared non‐breeding partial migration’); or (C) all individuals typically inhabit a non‐breeding location but some individuals sporadically migrate to breed at a different location during the breeding season while other individuals remain resident and hence do not breed (‘intermittent breeding partial migration’, also known as ‘skipped breeding partial migration’); or (D) all individuals typically inhabit a breeding location but some individuals sporadically migrate to a non‐breeding location during the breeding season and hence do not breed (‘intermittent non‐breeding partial migration’). Box sizes indicate local seasonal population densities, implying that density is highest when residents and migrants coexist. Background stippling indicates location–seasons where breeding can occur. Dark‐grey and light‐grey shading respectively indicate sets of individuals that do and do not breed in each season. These scenarios implicitly assume local strong seasonality such that: the migrants' non‐breeding‐season location cannot support breeding (A); the migrants' breeding location cannot support non‐breeding‐season survival (B, C); and the migrants' breeding‐season location cannot support breeding (D). Scenario C also requires an initial movement of offspring from the breeding location to the non‐breeding location.

Figure 3

Illustration of a simple partially migratory meta‐population (PMMP) system comprising four patches, where three patches (A–C) can support breeding‐season survival and reproduction and three patches (B–D) can support non‐breeding‐season survival. Here, individuals that breed in patch B (blue font) can remain resident (R) throughout both breeding and non‐breeding seasons or migrate (M, dashed arrows) to patches C or D for the non‐breeding season. Likewise, individuals that breed in patch C (yellow font) can remain resident or migrate to patches B or D for the non‐breeding season. Individuals that breed in patch A (red font) must migrate to patches B–D for the non‐breeding season. Migration could be bidirectional (e.g. patch C to patches B and D), or reciprocal (e.g. in both directions between patches B and C), but asymmetric (font sizes denote relative numbers of individuals). Patches B and C can consequently hold different combinations of resident and migrant individuals in both seasons (left versus right panels). Meanwhile, patches A and D are unoccupied in the non‐breeding and breeding seasons, respectively (i.e. local populations go seasonally extinct), yet support migrants in the opposite seasons. Patches A–C that can support breeding can also be linked by dispersal (solid black arrows). This general PMMP system thereby comprises a set of locations experiencing spatio‐temporal seasonal environmental variation that can be occupied by different sets of resident, migrant and dispersed individuals in different seasons.

Figure 4

Illustrations of key forms of demographic structure and covariation that could arise in partially migratory meta‐populations. (A) Migrants that breed in location 1 and spend the non‐breeding season in location 2 (red font) could have different reproductive success (RS_M) and breeding‐season survival (Φ_B,M) from seasonally sympatric residents in location 1 (blue font, RS_R1, Φ_B,R1), and from seasonally allopatric residents in location 2 (grey font, RS_R2, Φ_B,R2). These migrants could then have different non‐breeding‐season survival (Φ_NB,M) from seasonally sympatric residents in location 2 (Φ_NB,R2), and from seasonally allopatric residents in location 1 (Φ_NB,R1), creating additional structure in key demographic rates in both seasons. (B) An individual's form of migration versus residence (M vs R) might affect its reproductive success (RS), which might feed back to affect its subsequent migration or residence and resulting survival (Φ). (C) Migrants (M) that breed in location 1 might have higher reproductive success or seasonal survival than local residents (R), while migrants that breed in location 2 might have lower reproductive success or seasonal survival than local residents, creating spatially disruptive selection on migration. (D) Covariances between dispersal and migration. Covariance could be positive (left panel), where individuals i that disperse (solid arrow) from their natal location (dark grey) to a different breeding location (mid grey) are more likely to migrate (red dashed arrow) to a different non‐breeding‐season location (light grey) than individuals j that do not disperse. Conversely, covariance could be negative (right panel), where individuals i that disperse from their natal location are less likely to migrate than individuals j that do not disperse. (E) Complex feedbacks: the forms of migration versus residence (M vs R) and dispersal (Dis) could affect RS and Φ directly, and also indirectly if they affect each other (double arrows). Resulting RS could then feed back to affect the form of subsequent migration versus residence and dispersal (dotted arrows). (F) Example of non‐breeding‐season demographic structure arising if females are more likely to migrate (red font) from breeding areas (dark grey) to different non‐breeding areas (light grey) than males, while males are more likely to remain resident (blue font). Font sizes denote relative frequencies. An extreme environmental event in the non‐breeding area (black star) would then disproportionately impact migrant females. (G) Individual (a) residence (blue) or (b) migration (red) could be fixed and consistently expressed across multiple breeding and non‐breeding seasons, or could be plastic and expressed either (c) pre‐emptively or (d) responsively in some seasons but not others. Responsive migration in one season might lead to future pre‐emptive migration (d).

Figure 5

Summary of an evolutionary individual‐based model illustrating that multiple strategies of migration and residence can readily co‐exist across broad parameter space. (A) The modelled spatial structure comprised three zones that support reproduction (zone A), non‐breeding‐season survival (zone C) and both activities (zone B), with two patches (x and y) within each zone. Strategies of year‐round residence in B (BBB, black), and three forms of migration (dashed arrows: ACA, green; ABA, blue; and BCB, red), can all evolve. Dispersal can occur between x and y patches within A and B (solid arrows). (B) The number of indiviuals expressing each strategy after 20000 simulated generations of evolution varied with the survival cost of migration, but multiple strategies co‐existed over wide parameter space, generating multiple forms of partial migration. Data are from 500 independent simulations spanning a range of costs. (C) Snapshot of emerging spatio‐temporal variation in sub‐population size and composition, where pie and segment sizes respectively denote the total number of individuals, and the number of individuals expressing each strategy, that are present in the A, B, and C zones in the breeding and non‐breeding seasons. Black points indicate zone‐seasons with zero population. Example data are from one simulation, with cost of migration of 0.06 and strength of density‐dependence in non‐breeding‐season survival of 0.00015. (D) The proportion of individuals expressing each strategy after 20000 simulated generations also varied with the strength of density‐dependence in non‐breeding‐season survival, with an interaction with the survival cost of migration (probability of mortality during inter‐zone movement). Given high costs, population composition reduced to simple two‐location partial migration (segment a), or ‘leap‐frog’ migration (segment b), depending on the strength of density‐dependence. Data are from 20 replicate simulations for each combination of cost and density‐dependence. See Appendix S1 for details of the model.

Figure 6

Example time series from one simulation from an evolutionary individual‐based model illustrating population dynamics of individuals enacting three strategies of migration among zones A, B and C spanning three seasons (i.e. ABA, ACA and BCB), and of lifelong residence in zone B (i.e. BBB), and of total population size across 10000 simulated generations. Zone and movement structures are shown in Fig. 5A. See Appendix S1 for details of the model.