Discussion
The integration of genetic, morphological and ecological data, provide support to the hypothesis that even the barrier-free sea environment maintains a vast potential for ecological specialization. We delimited three independent clades of S. schlegeli by genomic data, and observed that this differentiation was mirrored in the morphological divergence of local populations. A similar phylogeographical pattern have been documented in the northwestern Pacific using fishes, molluscs, and crustaceans (Liu et al. 2007; Ni et al. 2012; Wilsonet al. 2020; Xu et al. 2009).
In marine environments, worldwide glaciation is confirmed to be the most efficient way in generating intraspecific genetic splits (Hewitt 2000), as well as shaping the present-day phylogeographical pattern of marine species (Dong et al. 2012; Wilson et al. 2020). Example of this have been found in different biogeographic realms, from the Indo-West Pacific to the Atlantic and Mediterranean basins (Cunhaet al. 2008; Ni et al. 2014). Pleistocene sea level fluctuation resulted in the isolation of marginal seas; for instance, sea level during the LGM decreased in the marginal seas of northwestern Pacific (approximately 100–120 m drop in the SCS, and 130–150 m drop in the ECS) (Wang & Sun 1994). Therefore, along with the sea level dropping, the ECS was reduced to an elongated trough, the Okinawa Trough, while the SCS became a semi-enclosed sac-shaped gulf connected with the Pacific mainly through the Bashi Strait (Wang & Sun 1994). These sea basins served as separate refuges for different geographic populations of local marine species. Clearly, the present genetic data suggests a relatively deep genetic split between the ECS-YS and SCS lineages, which has also been reported in other marine fishes (Liuet al. 2007). A few genomic regions in S. schlegeli showed high differentiation while the rest of the genome was very similar, as indicated by F ST analysis between North and SCS populations. Such a heterogeneous genome divergence has been commonly described in situations where differentiated genomes have experienced differential introgression following secondary contact (Xu et al.2009). Postglacial range expansions of marine species may bring formerly isolated populations into secondary contact at suture zones between seas including ECS/SCS (Ni et al. 2014). In fact, paleo habitat suitability projections showed a clear geographical separation between south and north populations during the LGM, which provide ecological support for the observed differentiation of ECS-YS and SCS lineages ofS. schlegeli . We hypothesize that Taiwan Strait land bridge may have served as a barrier to the dispersal of coastal marine species during the glacial periods, favouring inter-population genetic divergences.
Gene exchange among YS and ECS populations of S. schlegeliappears to be relatively frequent, as revealed by the genetic ancestry analyses, despite the fact that ECS and YS populations are clearly separated based on the PCA and phylogenetic tree clustering. During glacial periods, the northern regions of the YS were covered by land, and the ECS populations of S. schlegeli may have experienced expansion concomitant with a rise in sea level during interglacial periods, leading to genetic exchange between the ECS and YS populations (Chen et al. 2006; Yu & Zhou 2001). In parallel, genetic breaks may be attributed to freshwater outflow from rivers, as demonstrated, for example, in the Amazon basin (Rochaet al. 2002). As the third largest river in the world, Yangtze (Changjiang) River pours into the ECS with an average annual discharge of 924 billion m3 (Ni et al. 2012). While the barrier effect of the huge freshwater outflow on gene flow of coastal species is still a controversial issue, we suggest that the freshwater outflow may have affected the gene change of marine species in a more recent historical scale following potential changes in the estuary and variable river discharges. Given the fact that the freshwater outflow seems to affect mostly intertidal species with a strong habitat specificity (Dong et al. 2012), and SDM projections suggest that a large part of coastal areas of China, Japan, and Korea are suitable for S. schlegeli , we are inclined to believe that the phylogeographical pattern of S. schlegeli is largely explained by the historical glaciation.
It is worth noting that pipefishes are unique among fishes in that the males have a brood pouch and are ovoviviparous. Moreover, they are generally considered to be slow swimmers. These life history traits render them sensitive to environmental change (Vincent et al.2011; Wilson et al. 2003; Zhang et al.2017). Compared with other marine fishes, environmental heterogeneity is considered more likely to affect the S. schlegeli population structure due to local adaptation and the limited gene flow between different geographical populations. Hypervolume analysis indicated thatS. schlegeli from different geographic regions diverge in their realized niche, implying that the three clades have adapted to largely distinct environmental conditions including temperature and salinity. A recent study of Syngnathus typhle based on geometric morphometrics, prey availability, and dietary composition, demonstrated a pronounced variation in snout morphology across its distribution range, which may contribute to its adaptation in novel environments (Wilson et al. 2020). In the present study, a north-to-south variation of cranial morphology was detected in different S. schlegeli populations, which indicates that the individuals in the south of the distributional range (SCS) have smaller eye, and the opposite patterns are observed in the north (ECS and YS). The differences detected in the eye phenotype of pipefish from different populations may be related to differences in various environmental conditions, such as sea water turbidity, and food type and availability. Divergent selection resulted from the diverse environmental conditions could constitute barriers to gene flow (Tobler et al.2008), and identification of loci or genes under natural selection is important for determining the genetic basis of local adaptation affecting fitness traits (Zardi et al. 2013). A total number of 143 genes were annotated from the selected genome regions revealed by genome scan method, among which many key genes are involved in growth (pax7 ) (Akolkar et al. 2016), cold adaptation (csde1 ) (Yang et al. 2012), and eye development (eys , rx3 ) (Loosli et al. 2003; Yuet al. 2016). Accordingly, local populations of S. schlegeli may benefit from the adaptive variation to improve fitness of this species to local environments.
Despite the fact that marine fish generally show high genetic diversity and shallow population structure (Takahashi et al. 2015; Utter & Seeb 2010), the results of this study confirm that S. schlegeli has obvious geographical population structures along its distribution range in northwestern Pacific. The divergence among the different geographical populations may be related to the geological events, as well as adaptation to the habitat heterogeneity across the latitudinal gradients.