The presence of phylogenetically independent lineages adapting to similar environments offers great opportunities to investigate the roles of natural selection in phenotypic evolution . Furthermore, such replicate systems enable us to investigate the extent to which causative alleles and genes are shared among independent lineages adapting to similar environments and what factors determine the probabilities of sharing the same alleles and genes [2–5]. Such knowledge will help to understand the repeatability and predictability of evolution [2–5]. Because transition from marine to freshwater habitats occurred in multiple lineages [6, 7], we can find replicate pairs of closely related marine and freshwater organisms. Marine and freshwater environments differ in many biotic and abiotic factors. Therefore, phylogenetically independent lineages that achieved the marine–freshwater transition would offer great opportunities to investigate the genetic basis for parallel/convergent evolution accompanying freshwater colonization and adaptation [6, 7].
Among the organisms that have undergone the marine–freshwater transition, the three-spined stickleback (Gasterosteus aculeatus) are a remarkable system to study the genetic mechanisms underlying this transition [8–10]. The three-spined stickleback is a cold-water fish widely distributed in coastal marine, brackish, and freshwater habitats of the Northern hemisphere [11, 12]. Ancestral marine ecotypes of the three-spined stickleback colonized freshwater habitats across its distribution. Many of these habitats emerged following deglaciation during the Quaternary Period. Freshwater populations from different geographic regions often show similar morphology and physiology. Thus, the three-spined stickleback is an excellent system to investigate the genetic mechanisms underlying parallel evolution [11, 12].
Previous genetic studies on the parallel evolution of sticklebacks have mainly focused on postglacial freshwater populations in the Pacific Northwest of North America and in northern Europe [8–10]. The habitats in these regions were covered by ice sheets during the last glacial period and became uncovered within the last 12,000 years. Parallel evolution of several morphological and physiological traits in these postglacial populations has been caused by repeated fixation of identical-by-decent alleles [13–15]. Freshwater environments select freshwater-adaptive alleles that previously existed as standing variations in the founding populations or were carried from another freshwater population via gene flow. However, cases in which independent mutations of the same genes or different genes underlie parallel evolution have been described [9, 15–19]. Thus, it is crucial to understand the factors determining the use of shared and unique genes and alleles for parallel evolution to better understand the repeatability and predictability of evolution [2–5].
Recent studies have demonstrated that geographically distant lineages, such as East Pacific and Atlantic lineages, use different sets of standing genetic variations for parallel evolution [16, 19]. These results indicate that analysis of geographically diverse regions can help to understand the wide distribution of freshwater-adaptive alleles in G. aculeatus across its distribution [16, 19]. Such analyses can also clarify the alternative solutions when standing variations are not available [20, 21].
Japanese three-spined stickleback populations in the western Pacific basin offer several unique opportunities to investigate the genetic basis of parallel evolution (Fig. 1A). First, the Japanese Archipelago is geographically distant from North America and Europe, suggesting that the Japanese populations may share a relatively small number of genetic variants with North American and European populations. Previous studies have shown that reduction in the armor plate in freshwater populations in North America and Europe is caused by repeated fixation of the same ectodysplasin (Eda) allele, whereas armor plate reduction in Japanese freshwater populations is caused by independent mutations at Eda [8, 17, 20, 22].
Second, there are freshwater populations with different ages of colonization. The Japanese Archipelago was not covered by ice sheets in the Quaternary glaciation, suggesting that several freshwater habitats were accessible by sticklebacks well before 12,000 years ago. A previous mitochondrial DNA phylogenetic analysis estimated the divergence time of freshwater populations in Gifu and Shiga, central Honshu Island, termed ‘hariyo stickleback’ in Japan [23, 24], from the rest of G. aculeatus as 0.37–0.43 million years before present (Ma BP) based on a molecular clock. Additionally, there are several young freshwater populations. The three-spined stickleback inhabits lakes and ponds that were formed within 2,000–3,000 years in eastern Hokkaido. These freshwater populations are not genetically differentiated from marine G. aculeatus at allozyme or microsatellite loci [25, 26]. Several human-introduced populations also offer opportunities to investigate the genetic basis of rapid adaptation [27, 28]. Freshwater populations with such a diverse array of colonization ages provide opportunities to investigate how freshwater adaptation progresses over time.
Finally, the distribution range of G. aculeatus overlaps with that of its sister species G. nipponicus in northern Japan [29, 30]. Previous studies have shown that all freshwater populations examined thus far belong to G. aculeatus rather than G. nipponicus [18, 31]. G. aculeatus has higher copy numbers of the metabolic gene Fads2 and can survive better on freshwater-derived diets than G. nipponicus . Because there is past and ongoing hybridization between these two species [28, 32–34], it is important to determine the extent of introgression of freshwater-adaptive alleles between these two species to understand the genetic factors constraining the freshwater colonization of G. nipponicus.
As a first step towards a comprehensive understanding of the genetic basis of parallel evolution in the Japanese freshwater populations of Gasterosteus, we investigated their origins using phylogenomic approaches. The majority of previous phylogenetic studies on Japanese sticklebacks have used allozyme, microsatellite, and mitochondrial DNA. Mitochondrial DNA has been shown to introgress from G. nipponicus to G. aculeatus, suggesting that phylogeny based on mitochondrial DNA does not reflect the population history [33, 35, 36]. Previous phylogenetic analyses using allozyme and microsatellite were based on a small number of markers. More precise phylogenetic analysis with a large number of genome-wide single nucleotide polymorphisms (SNPs) are necessary. We have conducted phylogenetic analyses using whole genome sequences  and Restriction-site associated DNA (RAD) markers . However, we have identified several new habitats since then. Additionally, previous studies did not investigate the divergence time or phylogenetic relationships of the Japanese populations with North American and European populations. To solve these unanswered questions, we conducted a phylogenomic analysis of Japanese stickleback populations using RAD sequencing.