Habitat Use, Movement Patterns, and Spatial Population Structure of Polyphenic Sunfish

Populations of individuals are often spatially structured such that phenotypes are nonrandomly distributed, instead reflecting adaptive phenotype-habitat associations (Manel et al., 2003). The spatial distribution of phenotypes can influence demographic (e.g. population growth rate) and evolutionary parameters (e.g. gene flow) that are important determinants of population dynamics (Manel et al., 2003) and adaptive evolutionary responses (Kinnison & Hairston, 2007). For example, the population dynamics of the Granville fritillary butterfly in Finland are governed by a balance between stochastic local extinctions and recolonization through a network of populations, or a metapopulation (Hanski et al., 1995). For this species, local recolonization and extinction risk are both highly dependent on the spatial arrangement of populations (Saccheri et al., 1998). Spatial population structure is similarly important for evolutionary processes like adaptive divergence, for example when the spatial distribution of phenotypes determines opportunities for assortative mating and competitive interactions (Rice, 1987; Caillaud & Via, 2000). Because of its importance for both ecology and evolution, it is critical to understand how spatial population structure is generated and maintained under various scenarios of environmental heterogeneity. Here, I investigated the spatial population structure of ecologically and phenotypically diverse pumpkinseed sunfish (Centrarchidae: Lepomis gibbosus) within a single lake, specifically extending our characterization of phenotype-habitat associations to an understudied habitat type, and quantifying connectivity between habitat patches over one year. Adaptive phenotype-habitat associations can occur across a range of scales from microhabitat use within populations to adaptive divergence among populations. I consider a population to be a group of conspecific individuals that is spatially distinct from other groups of individuals where members of the group can potentially interbreed (Wells & Richmond, 1995). A metapopulation is a set of populations with some amount of migration between breeding seasons (Wells & Richmond, 1995). Ecologists have a long history of studying the spatial structure of metapopulations and the effects of migration on local population dynamics and persistence, but evolutionary biologists also acknowledge these different scales when they distinguish processes regulating adaptive divergence in sympatry (within populations) and allopatry (among populations). Geographic scale often informs our view of population structure. Here, I focus on fine scale spatial population structure within a single lake population of fish because of its relevance to competitive and reproductive interactions. In this thesis, I studied the spatial population structure of sunfish within a single lake roughly 2.59 km2 in area. I consider a whole lake to be an environment containing a set of distinct habitat patches. Habitat patches are identified by spatially discrete clusters of reproductive activity that are separated by some geographic barrier where no reproduction occurs, for example deep open water, or a rocky point separating shallow bays. Spatial population structure is the product of environmental heterogeneity, the variation in abiotic (e.g. temperature) and biotic (e.g. predator assemblage) conditions between habitat patches, and the dispersal capacities of individuals (Hanski, 2012). The environment occupied by a population is rarely homogenous but is instead made up of multiple habitat patches of varying type and quality. Differences in habitat conditions can result in different locally optimal phenotypes, such as when ecological performance involves strong trade-offs that preclude ecological generalists (Futuyma & Moreno, 1988). With sufficient genetic variation, diversifying selection for alternate optimal phenotypes can lead to local adaptation (Rainey & Travisano, 1998). A useful example can be found in populations of Timema cristinae walking stick insects that occupy different habitat patches characterized by a single shrub type and the patch size (Sandoval, 1994). Each shrub type has a different colour background and so has an optimal colour phenotype that allows camouflage from predators, while the size of the patch determines the local carrying capacity (Sandoval, 1994; Nosil, Crespi & Sandoval, 2003). A predictable spatial distribution of phenotypes across habitat patches emerges in this system that is generated by the pattern of environmental heterogeneity the population experiences (Sandoval, 1994; Nosil, Crespi & Sandoval, 2003). However, the effect of environmental heterogeneity is dependent upon its spatial scale relative to the dispersal capacities of the individuals, or the ‘connectivity’ of a population. Spatial population structure can be homogenized or maintained by the patterns of connectivity between habitat patches (Garcia-ramos & Kirkpatrick, 1997; Nosil & Crespi, 2004; Hanski, 2012). Connectivity is a combination of the physical properties of paths between patches and the dispersal capacity of individuals. It links habitat patches demographically and genetically through the movement of individuals and their genes (Hendry & Taylor, 2004; Hanski, 2012). Connectivity has important population consequences when differences in quality between habitat patches results in local populations with positive growth rates that are net exporters of migrants (sources), or with negative growth rates where they can only be maintained by receiving migrants (sinks) (Pulliam, 1988; Dias, 1996). From an evolutionary perspective, source populations that produce migrants can create a migration load in sink populations, as immigrants bring non-adaptive genes, preventing local adaptation (Bolnick & Nosil, 2007; Bolnick, Caldera & Matthews, 2008). Populations of blue tit (Parus caeruleus) in southern France provide an informative case of source-sink dynamics, where birds occupy habitat patches with either deciduous or sclerophyllous forest types (Dias & Blondel, 1996). Sclerophyllous habitat patches are sinks because birds there have mismatched breeding phenology that prevents exploitation of peak caterpillar abundance. Conversely, deciduous habitat patches are sources as birds there are well matched to prey population dynamics. Local adaptation in sclerophyllous sink habitat patches is prevented because they receive a high migration load from the deciduous source habitat patches which spatially dominate the environment (Dias & Blondel, 1996). Despite the variation in conditions between deciduous and sclerophyllous habitat patches, there is a more uniform spatial distribution of phenotypes. In this case, connectivity between habitat patches degrades spatial structuring of blue tit populations. Models examining different scenarios of environmental heterogeneity and connectivity find that three basic patterns of spatial phenotype distributions can emerge; locally adapted specialists in each habitat patch, a single habitat generalist across all habitat patches, or a single dominant habitat specialist across all habitat patches (Hanski & Mononen, 2011). Generally, higher habitat heterogeneity (i.e. larger ecological differences between habitat patches) promotes local adaptation while higher connectivity breaks down spatial population structure. An exception to this rule is when connectivity between habitat patches is additionally influenced by matching habitat choice (Edelaar, Siepielski & Clobert, 2008; Scheiner, 2016). When individuals can match themselves to the habitat type where they are best adapted, then connectivity can allow a positive feedback between local adaptation and reduced migration that promotes spatial population structure and adaptive divergence (Ravigne, Dieckmann & Olivieri, 2009; Scheiner, 2016) Beyond being a determining factor in population persistence, spatial population structure also can influence community level processes (McCann, Rasmussen & Umbanhowar, 2005). The amount of connectivity between habitat patches and local adaptation within habitat patches will influence how interacting populations move energy through food webs (Bolnick et al., 2003; McCann, Rasmussen & Umbanhowar, 2005). For instance, energy flow between habitat patches will tend to be limited in a population structured into distinct clusters of locally adapted phenotypes (Quevedo, Svanbäck & Eklov, 2009; Svanback et al., 2015). Alternatively, if a population has high connectivity between habitat patches or less local adaptation, energy flow between habitat patches will be enhanced (Vander Zanden & Vadeboncoeur, 2002; McCann, Rasmussen & Umbanhowar, 2005). Ultimately, some level of connectivity can affect whole food web stability and productivity by stabilizing species interactions, although too much connectivity can be destabilizing (Rooney & McCann, 2012). Despite the importance of understanding the origins and maintenance of spatial population structure, and a solid theoretical foundation, there are few case studies of how spatial population structure arises in natural settings (Hanski, 2012). Here, I investigated several questions about how spatial population structure is generated and maintained in an ecologically and phenotypically diverse (‘polyphenic’) population of pumpkinseed sunfish (Centrarchidae: Lepomis gibbosus). Here, I use polyphenic to refer to a population where phenotypic diversity has a significant plastic component as opposed to a polymorphic population where phenotypic diversity is primarily the result of genetic diversity. Pumpkinseed sunfish are a widespread freshwater fish found throughout eastern North America, with the most northerly distribution of the Lepomis sunfishes. Typically, they exploit the littoral (near shore, benthic) habitat, but in low productivity lakes with limited littoral habitat and no competitors for zooplankton prey (e.g. bluegill sunfish, L. machrochirus), pumpkinseed will also exploit the pelagic (open water, limnetic) habitat (Robinson, Wilson & Margosian, 2000). These lake habitats differ in abiotic conditions such as physical complexity and water clarity, and biotic conditions such as prey availability and predator risk. For example, littoral habitat has a greater abundance and diversity of benthic macroinvertebrates like snails and insect larvae, while pelagic habitat has a greater abundance and diversity of zooplankton (Robinson et al., 1993; Jastrebski & Robinson, 2004). These polyphenic sunfish present an ideal opportunity to address questions about how spatial population structure is generated and maintained. There are predictable associations between phenotypes and habitat type within polyphenic populations of pumpkinseed sunfish, characteristic of spatially structured populations. There are differences in functional foraging traits between sunfish from littoral and pelagic habitats, including body shape, pharyngeal jaw size, and gill raker density (Robinson et al., 1993; Gillespie & Fox, 2003; Colborne et al., 2016). These morphological differences characterize different phenotypes within the population that are associated with a specific ecology, or ‘ecotype’. Sunfish likely face diversifying selection for these alternate phenotypes as morphological differences are related to foraging performance on habitat-specific prey types (Parsons & Robinson, 2007) and components of fitness like growth and body condition (Robinson & Wilson, 1996; Jastrebski, 2001; Colborne et al., 2016). Genetic variation underlies some of these phenotypic differences between ecotypes, but these traits also have strong plastic developmental responses to local conditions that likely contribute to phenotype-habitat associations (Robinson & Wilson, 1996; Parsons & Robinson, 2007). This general pattern of phenotype-habitat association is repeated in multiple sunfish populations in post-glacial lakes in northeastern North America (Weese, Ferguson & Robinson, 2012). We do not fully understand how connectivity between habitat patches shapes population structure within these populations. Within a summer growing season, sunfish ecotypes exhibit strong patch fidelity, segregating between littoral and pelagic habitat patches for foraging and reproduction (McCairns & Fox, 2004). A mark-recapture study of sunfish during the summer found 97% and 98% probability of site fidelity for pelagic and littoral individuals, respectively (McCairns & Fox, 2004). Additionally, among sunfish transplanted between habitat patches, there was a 74% and 93% probability of returning to their site of capture for pelagic and littoral individuals, respectively (McCairns & Fox, 2004). However, sunfish live up to 10 years, and they leave summer habitats patches in the fall to overwinter in deeper waters and return to summer season sites in the spring (personal observation). It is unclear how much they may move between habitat patches between years. Genetic differentiation at neutral loci is low or nonexistent in all polyphenic sunfish populations, which suggests substantial gene flow between ecotypes (Weese, Ferguson & Robinson, 2012; Colborne et al., 2016). A major goal of this thesis is to improve our understanding of connectivity in these populations and how this may influence population structure. In chapter 1, I explored spatial population structure in a polyphenic population of pumpkinseed sunfish, specifically quantifying how a previously unstudied habitat, exposed lake shorelines, influences spatial population structure. This builds on our current understanding of spatial structuring across the littoral-pelagic habitat divergence to include other frequently used habitat types in post-glacial lakes. In chapter 2, I used a mark-recapture study to estimate between-year rates of adult sunfish movement and to understand how connectivity between habitat patches influences spatial population structure. This specifically extends our understanding of connectivity in these populations to a scale that has not been previously studied. The findings of this thesis increase our understanding of how environmental heterogeneity and connectivity interact to generate spatial population structure in populations of sunfish. The perspectives gained will further the long-term goal of identifying the factors that regulate adaptive divergence in postglacial lake fishes (Robinson & Wilson, 1994; Robinson, Wilson & Margosian, 2000).