Introduction
In eukaryotes, adaptation of populations to novel ecological conditions often occurs from standing genetic variation (SGV), that is, selectively relevant variation pre-existing in the ancestor (Orr & Betancourt 2001; Hermisson & Pennings 2005; Barrett & Schluter 2008; Messer & Petrov 2013; Matuszewski et al. 2015). A puzzle, however, is how SGV is maintained in the ancestor (Yeaman 2015): if genetic variants are favored by selection in a novel, derived habitat, should they not be unfavorable and hence eliminated by purifying selection in the ancestral habitat? One solution to this paradox is that genetic variants favored in the derived habitat are maintained as SGV in the ancestor by continued hybridization (and hence gene flow) between derived and ancestral populations, thus counteracting the selective removal of these variants in the latter (Colosimo et al. 2005; Bolnick & Nosil 2007; Barrett & Schluter 2008; Schluter & Conte 2009; Yeaman & Whitlock 2011; Galloway et al. 2020). An alternative idea is that variants beneficial within the novel habitat are selectively neutral in the ancestral population when their frequency is relatively low. While this must obviously hold for recessive variants (Barrett & Schluter 2008), quantitative genetic models suggest that when the traits under selection are highly polygenic (that is, influenced by a great number of loci), adaptive divergence may generally occur primarily via the establishment of linkage disequilibrium among alleles and involve only relatively subtle (or at least incomplete) allele frequency differentiation (Latta 1998; Kremer & Le Corre 2012; Le Corre & Kremer 2012). In this case, SGV could persist in the ancestor simply because there is no purifying selection to complete its elimination. The relative importance of these two not mutually exclusive explanations for the maintenance of SGV, gene flow-selection balance and selective neutrality, remains unknown and has, to the best of our knowledge, not been subject to empirical investigation. An obstacle for doing so is that organismal systems are required in which adaptive genetic variation can be detected and quantified in both derived and ancestral populations simultaneously.
We here perform such an investigation in threespine stickleback fish (Gasterosteus aculeatus ) by focusing on genetic variation promoting the adaptation of populations to acidic freshwater habitats after the recent (postglacial) colonization of these habitats by ancestral marine stickleback. Adaptation to acidic waters likely involves numerous traits, but particularly obvious elements include the reduction of external skeletal armor and body size in some acid-adapted stickleback populations relative to their ancestor (and to standard freshwater-adapted stickleback) (Figure 1a) (Campbell 1985; Giles 1983, Bourgeois et al. 1994; Spence et al. 2013; Klepacker et al. 2016; Magalhaes et al. 2016; Haenel et al. 2019). The function of this evolution is likely reduced metabolic demands, conferring an advantage in nutrient-depleted acidic habitats. (Note that for simplicity, we will use the terms acidic habitats and acidic adaptation throughout this paper, but we acknowledge that selection may not necessarily be mediated by pH (alone), but by an associated shortage in dissolved ions). Although marine threespine stickleback have colonized innumerable freshwater habitats across the northern hemisphere, morphological adaptation to acidic habitats is reported from relatively few locations across the species’ range only (Campbell 1985; Bourgeois et al. 1994; Klepaker et al. 2013). An exception is North Uist (Outer Hebrides, Scotland) (Figure 1b), an island on which acidic-adapted stickleback ecomorphs are common. Due to its particular surface geology (Waterston et al. 1979), the eastern part of this island harbors numerous acidic lakes (pH around 5-6) inhabited by archetypal acidic-adapted stickleback that have likely evolved multiple times independently (Giles 1983; Spence et al. 2013; Klepacker et al. 2016; Magalhaes et al. 2016; Haenel et al. 2019). This parallel evolution has occurred though the deterministic sorting of SGV available in the marine ancestor, because alleles recruited repeatedly for acidic adaptation are consistently found in extant marine stickleback breeding in coastal habitats of North Uist, albeit generally at modest to low frequency (Haenel et al. 2019). What remains unknown is whether this SGV primarily reflects the continued flow of acid-favored alleles into marine stickleback by hybridization, or whether alleles beneficial to acidic adaptation segregate largely neutrally in marine fish.
To address this question, we here use whole-genome sequence data to examine SGV in marine stickleback across the Atlantic Ocean. We hypothesize that if the presence of SGV relevant to acidic adaptation in marine stickleback around North Uist reflects a balance between gene flow and purifying selection, the frequency of alleles favored in acidic habitats should be elevated in marine stickleback breeding around North Uist compared to marine stickleback sampled from more distant locations. The reason is that acidic lakes represent an uncommon freshwater habitat outside North Uist, and the acidic-adapted ecomorphs common on this island are rare on a worldwide basis. Purifying selection should therefore vastly outbalance the input of deleterious acidic-favored alleles by hybridization in marine stickleback far from North Uist. Alternatively, the frequency of acidic-favored alleles may not be elevated in marine stickleback breeding around North Uist compared to marine fish in general, suggesting that purifying selection against these alleles is weak or absent in marine stickleback at large. As we show, our data support this latter scenario, thus highlighting selective neutrality as an underappreciated explanation for the maintenance of SGV.