Introduction

Understanding how eco-evolutionary processes and environmental factors drive population differentiation and adaptation remain key challenges in evolutionary biology. Besides informing about the origin and dynamics of biodiversity, knowledge about drivers can promote successful management and protection of populations and species. For differentiation to occur populations must be at least partially reproductively separated. That geographically isolated (allopatric) populations experience such reproductive separation is self-evident. However, reproductive isolation may also arise through other mechanisms that restrict gene flow. For example, the Isolation By Distance model (IBD; Wright, 1943) proposes that there is a negative association between geographic distance and gene flow, which should translate into clinal population structures with higher degrees of differentiation on longer distances. Apart from geographical dispersal boundaries, ecological factors can influence the degree of gene flow (Rundle & Nosil, 2005). Differences in ecology and environmental preferences (e.g.different ecotypes) can instigate population differentiation via Isolation By Environment (IBE; Sexton, Hangartner, & Hoffmann, 2014; Wang & Bradburd, 2014). There is also potential for population genetic structure to evolve as a result of Isolation By Time (IBT), whereby reproductive separation occurs between groups of individuals due to differences in reproductive timing (Hendry & Day, 2005).
Differences in ecology and/or environmental preferences are generally accompanied by divergent selection. The evolution of different local adaptations can cause a feed-back loop that reinforces the reproductive separation and speeds up the differentiation process (a case of IBE often referred to as Isolation by Adaptation (IBA); Nosil, Egan, & Funk, 2008). The reinforcing effect is likely of particular importance when such divergent selection acts upon traits directly connected to reproductive isolation and reproductive success, e.g. habitat preferences (Hendry, Nosil, & Rieseberg, 2007), spawning segregation (Nosil, 2012), or early life history traits (Momigliano et al., 2017), and it might ultimately lead to ‘ecological speciation’ (Schluter & Rambaut, 1996).
Northern temperate freshwater fish species present good opportunities to study differentiation and eco-evolutionary dynamics (Hume, Recknagel, Bean, Adams, & Mable, 2018). It was not until after the retreat of the glaciers (15,000 – 10,000 years ago) that many areas were colonized via range expansions (Petit et al., 2003; Rowe, Heske, Brown, & Paige, 2004). The geographical isolation in glacial refugia, in combination with the subsequent range expansions, means that there has been opportunities for founder events, bottlenecks, divergent selection, and adaptions to local conditions to influence genetic structure and diversity, and the different contemporary ecotypes that can be observed today have likely evolved during this time. In aquatic systems, gene flow is generally expected to be high due to the lack of apparent dispersal boundaries (Gagnaire et al., 2015; Puebla, Bermingham, & McMillan, 2012). Despite this, genetic structuring has been reported within open and connected waterbodies (Bergek & Björklund, 2007; Momigliano et al., 2017; Nordahl et al., 2019; Rosenbaum et al., 2009; Willing et al., 2010; ), even on small spatial scales, which is likely due to ecological and environmental heterogeneity imposing limits to gene flow. One example of such a system is the Baltic Sea, one of the largest brackish-water ecosystems in the world, that exhibits steep environmental north-south gradients in salinity and temperature (Bendtsen, Söderkvist, Dahl, Hansen, & Reker, 2007). This makes the Baltic Sea an excellent system for studies of differentiation and local adaptation (e.g. in Guo, DeFaveri, Sotelo, Nair, & Merilä, 2015; Guo, Li, & Merilä, 2016).
One of the most common, coastal, predatory fish species in the Baltic Sea is pike (Esox lucius ). Pike plays an important role in many aquatic ecosystems by regulating abundances of species in lower trophic levels (Donadi et al., 2017). It is an economically important species for both recreational and commercial fishing (Lehtonen, Leskinen, Selen, & Reinikainen, 2009; Pierce, Tomcko, & Schupp, 1995), and an established model species in studies of ecology and evolution (Forsman et al., 2015). It is originally a freshwater species, but it has managed to colonise brackish waters with salinities up to approximately 15 ppt (Craig 2008). This range expansion has been accompanied by the evolution of three different ecotypes that differ with regard to migratory behaviour. The original ecotype (freshwater) spend the entire life in freshwater, whilst the other ecotypes (anadromous and resident) spend either part of, or their entire life, in saline (brackish) waters. The two latter ecotypes co-exist in the coastal waters of the Baltic Sea during most of the year (Westin & Limburg, 2002), and separate only for a short period during spawning when the anadromous individuals migrate to freshwater localities whilst the resident ecotype stays in the brackish coastal waters (Engstedt, Stenroth, Larsson, Ljunggren, & Elfman, 2010; Larsson et al., 2015; Muller, 1986). An important feature of pike in the Baltic Sea, which affects the population structure of the species (Nordahl et al., 2019), is their homing behaviour (Engstedt, Engkvist, & Larsson, 2014; Larsson et al., 2015; Tibblin, Forsman, Borger, & Larsson, 2016), i.e. the adults return to their natal spawning ground for reproduction (Engstedt et al., 2014; Engstedt et al., 2010; Jacobsen et al., 2017; Muller, 1986).
While a previous study of pike in the Baltic Sea has indicated genetic differentiation between the resident and anadromous ecotypes (Nordahl et al., 2019), studies of the genetic structuring within the ecotypes show conflicting results. Some studies suggest weak structuring (Laikre et al., 2005; Wennerström, Olsson, Ryman, & Laikre, 2016), whilst others report fine-scaled genetic structuring among anadromous populations, and point to a role of isolation by distance (Möller, Winkler, Richter, & Bastrop, 2020; Nordahl et al., 2019; Sunde, Yildirim, Tibblin, & Forsman, 2020a). There is also evidence to suggest that environmental differences among spawning locations have resulted in adaptive phenotypic differentiation in various traits, including salinity tolerance (Sunde, Tamario, Tibblin, Larsson, & Forsman, 2018), temperature tolerance (Sunde, Larsson, & Forsman, 2019), vertebral number (Tibblin, Berggren, Nordahl, Larsson, & Forsman, 2016), body size and growth rate (Tibblin et al., 2015), early life history traits, and reproductive investment (Berggren, Nordahl, Tibblin, Larsson, & Forsman, 2016). Pike in the Baltic Sea therefore offers possibilities to study differentiation at different levels (between allopatric and sympatric populations, and within and among ecotypes), to evaluate the contributions to genetic structure of different types of reproductive isolation (such as IBD, IBE, IBA, and IBT), and to investigate the potential genetic underpinnings of local adaptations.
Increased knowledge into these issues is also important for management. Declines in pike populations in the Baltic Sea have been observed during the last decades (Lehtonen et al., 2009; Ljunggren et al., 2010; Nilsson, Flink, & Tibblin, 2019; Olsson, 2019), and extensive management actions, such as restoration of spawning habitats and fisheries regualtions, have been implemeted to counteract this negative trend (Larsson et al., 2015; Nilsson et al., 2019). Despite this, abundances remain low, and there is a need to increase the understanding to aid successful management. Previous population assignments have mainly been based on studies utilizing neutral microsatellite markers (Bekkevold, Jacobsen, Hemmer‐Hansen, Berg, & Skov, 2015; Eschbach et al., 2019; Möller et al., 2020; Nordahl et al., 2019; Wennerström et al., 2016; but see Sunde et al., 2020b). Informed decisions regarding management and conservation efforts require knowledge about how the combination of stochastic and deterministic processes shape genetic structure and diversity. This in turn requires that population genetic studies use molecular markers that capture variation also in coding regions (Sunde et al., 2020a). To investigate genetic structure and decipher the roles of different ecological and evolutionary processes for differentiation and adaptation in pike, we therefore performed a population genetic study utilizing Restriction Site Associated DNA sequencing (RADseq). For this we used 11 populations spanning 8.7 degrees in latitude, from Denmark (Askeby) in the south to Umeå in northern Sweden (Figure 1, Table 1 ), that experience different environmental conditions (e.g. ice cover, light, salinity, temperature, prey species, competition, and predation). The populations represented all three ecotypes (freshwater, anadromous and resident) and, to our knowledge, this is the first population genetic study of pike that includes populations of all three ecotypes.
The RADseq method yields thousands of Single Nucleotide Polymorphisms (SNPs) residing in both coding (functional) and non-coding (mainly selectively neutral) genomic regions (Andrews, Good, Miller, Luikart, & Hohenlohe, 2016). Thus, comparisons of genetic structure and diversity obtained using separate analyses of neutral and functional SNPs, can inform about the roles of both stochastic and deterministic processes (Andrews et al., 2016; de Villemereuil & Gaggiotti, 2015; ). In the present study, we therefore investigated genetic structure using both neutral and adaptive SNP datasets. In addition, we performed outlier analyses to identify loci putatively under selection and to pinpoint environmental factors contributing to evolutionary divergence among ecotypes and populations.