Hidden diversity and phylogeographic history provide conservation insights for the edible seaweed Sargassum fusiforme in the Northwest Pacific

Abstract Understanding the evolutionary processes that have created diversity and the genetic potential of species to adapt to environmental change is an important premise for biodiversity conservation. Herein, we used mitochondrial trnW‐L and cox3 and plastid rbcL‐S data sets to analyze population genetic variation and phylogeographic history of the brown alga Sargassum fusiforme, whose natural resource has been largely exterminated in the Asia–Northwest Pacific in the past decades. Phylogenetic trees and network analysis consistently revealed three major haplotype groups (A, B, and C) in S. fusiforme, with A and B distributed in the Japan‐Pacific coast. Group C consisted of three subgroups (C1, C2, and C3) which were distributed in the Sea of Japan, the Yellow–Bohai Sea, and East China Sea, respectively. Isolation‐with‐migration (IM a) analysis revealed that the three groups diverged approximately during the mid‐Pleistocene (c. 756–1,224 ka). Extended Bayesian skyline plots (EBSP) showed that groups A and B underwent relatively long‐term stable population size despite a subsequent rapid demographic expansion, while subgroups C2 and C3 underwent a sudden expansion at c. 260 ka. FST and AMOVA detected low population‐level genetic variation and high degrees of divergence between groups. The cryptic diversity and phylogeographic patterns found in S. fusiforme not only are essential to understand how environmental shifts and evolutionary processes shaped diversity and distribution of coastal seaweeds but also provide additional insights for conserving and managing seaweed resources and facilitate predictions of their responses to future climate change and habitat loss.


| INTRODUCTION
Population genetic differentiation and biogeographic regionalization, as the central issues in conservation biology, have been well illustrated to directly relate to evolutionary forces such as isolation, mutation, and historical demography (Dufresnes et al., 2013;Millar & Byrne, 2013). Thus, understanding phylogeographic processes that determine genetic diversity and distribution patterns of lineages over time and space has become a fundamental goal of conservation biogeography (Richardson & Whittaker, 2010). On the one hand, phylogeographic lineages within species constitute ecologically and evolutionarily significant units (ESUs) for conservation and management (D'Amen, Zimmermann, & Pearman, 2013;Dufresnes et al., 2013). On the other hand, evolutionary processes at either macro-or microgeographic scale can recreate, maintain, and restore specific phenotypic or genotypic variants-the priority goals of conservation planning (Crandall, Bininda-Emonds, Mace, & Wayne, 2000;Moritz, 1999). Changing climate may impact species ranges, biodiversity, and evolutionary legacy (Lavergne, Mouquet, Thuiller, & Ronce, 2010). Therefore, it is crucial to characterize the genetic lineages that may reflect the adaptive potential of species and may be relevant to future response to environmental change.
The Asia-Northwest Pacific (ANP) is an important marine center harboring rich seaweed diversity and endemism in the world's oceans (Keith, Kerswell, & Connolly, 2014;Kerswell, 2006). On the coasts of mainland East Asia, there are several well-collected regions (e.g., China and Korea) where around 500 species of seaweed have been recorded (Norton, Melkonian, & Andersen, 1996;Tseng, 1983). In the Japanese Archipelago, similar collecting efforts have documented 1,510 seaweed species which are two to three times more than those in regions with similar latitude ranges (e.g., S. Alaska-Oregon and N. British Isles-Morocco) (Norton et al., 1996). However, the distribution pattern of seaweed richness in the ANP, particularly around Japan and China, has been largely destroyed by environmental shifts in the past few decades (Sun, Ning, Le, Chen, & Zhuang, 2010;Tanaka, Taino, Haraguchi, Prendergast, & Kiraoka, 2012). For instance, the Nanji Island, China, underwent a rise in annual sea surface temperature (SST) by 0. 5°C during 1960-2006, with the species number of cold-temperate seaweeds reduced from 7 in 1959 to 2 in 2006 (Sun et al., 2010). The rapidly decreased seaweed diversity imposes an urgent need to manage and conserve seaweed resources and biodiversity under accelerating environmental changes.
Sargassum fusiforme (Harvey) Setchell is a large brown alga found in the lower intertidal zone throughout southern Japan, the rim of the Yellow-Bohai Sea, and East China Sea (Figure 1). Ecologically, this species can form dense beds and make substantial contribution to coastal marine ecosystems by acting as primary producer, providing spawning, nursery, and feeding grounds for fish and invertebrates, and improving environmental conditions (e.g., water motion and temperature) (Hwang, Yoo, Baek, & Park, 2015). Socioeconomically, S. fusiforme is widely utilized as food, polysaccharide resource, and medical agent (Mao, Li, Gu, Fang, & Xing, 2004;Zhu, Heo, & Row, 2010). Its market value in Japan was estimated to be 130 billion JPY during the late 1990s (Murata & Nakazoe, 2001). The large market demand caused extensive harvest of S. fusiforme in nature, with an annual production of 8,000-10,000 tons in Japan (Fresh weight) (Ito, 2012) and 32,000 tons in China in 2007 (Pang, Shan, Zhang, & Sun, 2008). Profit-induced catastrophic harvest, together with habitat degradation, resulted in F I G U R E 1 Haplotype distribution pattern (a) and maximum-likelihood (ML, upper)/Bayesian inference (BI, lower) (b) inferred from mtDNA trnW-L+cox3 data. Shaded sea areas are continental shelves that would have been exposed to the air during periods of low sea level. Numbers at tree nodes are percentages of bootstrap support. Each line between main haplotypes represents one mutation step. Detailed locality information is shown in annual production declined from 559 to 7.74 tons (Zhang & Liu, 2009).
Apart from commercial harvest and habitat loss, long-term changes in regional SST also have generated serious impacts on S. fusiforme on the coast of southern Japan and China. It is estimated that the SST rose by 1-2°C along Kagoshima, Japan, in the past four decades (Tsuchiya, Sakaguchi, & Terada, 2011) and the average SST of Kyushu Island increased by 1.2°C during 1900-2010 (Japan Meteorological Agency 2011), leading to a massive reduction in the distribution range and biomass of S. fusiforme at marginal areas (Kokubu et al., 2015). In Nanji Island, the rising SST caused S. fusiforme-S. thunbergii (Mertens ex Roth) Kuntze codominated beds in zonal community in 1959 to become dominated solely by S. thunbergii in 2006 (Sun et al., 2010). The contraction of the distribution range and loss of production of S. fusiforme in the ANP thus raise an essential question of how to practice efficient measures to conserve this commercially important seaweed species.
Although the characterization of intraspecific diversity and phylogeographic structure is fundamental to the conservation and management of species (Newton, Allnutt, Gillies, Lowe, & Ennos, 1999), a comprehensive attempt has yet to be carried out across the range of S. fusiforme. A recent molecular survey using mitochondrial cox1 revealed that oceanic currents drove asymmetric genetic exchange between S. fusiforme populations in the ANP (Hu, Zhang, Lopez-Bautista, & Duan, 2013), yet the cryptic lineage diversity and evolutionary patterns remain largely unresolved. From a conservation genetic perspective, the failure to survey population genetic structure of S. fusiforme may result in overexploitation or localized extirpation of uncharacterized biodiversity (Hueter, Heupel, Heist, & Keeney, 2005). Deciphering the pattern and degree of population subdivision and structured lineage diversity becomes a prerequisite for conserving and managing the S. fusiforme resource.
In this study, our main goals were as follows: (i) to quantify the phylogeographic structure and large-scale assessment of genetic variation within and between populations by integrating mitochondrial and plastid loci, (ii) to detect the historical demography and geographic distribution of lineage/group diversity in the natural range, and (iii) to place current patterns of genetic diversity and phylogeographic structure into both historical and conservation context with the aim of sustaining natural seaweed resources in the ANP.

| Molecular diversity, phylogeny, and principal component analysis (PCoA)
Sequences were aligned with MEGA 5.0 (Tamura et al., 2011). Genetic diversity was estimated by the number of polymorphic sites (S), haplotype distribution, number of haplotypes (Nh), haplotype diversity (h), and nucleotide diversity (π), which were all calculated in arleqUin 3.5 (Excoffier & Lischer, 2010). To evaluate the relationships among haplotypes, a parsimony median-joining network was generated with the program network 4.51 (Bandelt, Forster, & Röhl, 1999). JMODELTEST 0.1.1 (Posada, 2008)  components were tested using 10 5 permutations. Analysis of molecular variance (AMOVA) was also conducted in arleqUin to assess the spatial partitioning of genetic variance among the groups defined above.

| Dating divergence time
The isolation-with-migration model implemented in IMa2 (Hay & Nielsen, 2007;Nielsen & Wakeley, 2001) was used to estimate migration rates (m), population sizes (Θ), and divergence time (t) based on the concatenated trnW-L and cox3 data sets (running as independent loci because they exhibited substantially different mutation rates [Chan et al., 2013]). IMa can incorporate multiple genetic loci independently with locus-specific mutation scalars. One way to implement such scalars is to pick one locus as a standard with a mutation rate scalar of one and to have the scalars for other loci vary as parameters to be estimated. Specifically, IMa jointly estimates these genetic parameters in a Bayesian sampling framework by calculating posterior probabilities for parameters across a set of likely gene trees (Nielsen & Wakeley, 2001). Several runs were conducted until stationary results were achieved. The final runs included 100 coupled Markov chains, a burn-in period of 300,000 steps, and a geometric heating model, with the first and second heating parameters of 0.99 and 0.90 for individual chains, respectively. A total of 90,000 genealogies were sampled to estimate the joint posterior probability distributions of the migration parameters. It should be noted that here, we just tried to estimate the relative divergence times between genetic groups defined by phylogenetic analyses because mutation rates are time dependent over evolutionary timescales (Ho et al., 2011). In the IMa analysis, we set T A B L E 1 Diversity indices of Sargassum fusiforme populations inferred from mitochondrial TrnW-M+cox3 and plastid rbcL-S both the mutation scalars and mutation rate (the average value of mutation range of mtDNA Cox3, 0.0010-0.0016 site −1 Myr −1 [Chan et al., 2013]) in the input files. The generation time, g (1 year), was applied to convert the output into units of years.

Sampling localities (Abbreviation) Coordinates
The relative divergence time between groups was also calculated using the equation dA = 2μT, where μ is the average substitution rate per site of gene and dA is the net average genetic distance between groups (Nei & Li, 1979). The net genetic distance between groups was measured for multiple hits using the equation dA = dXY-(dX + dY)/2 with 1000 replicates in MEGA 5.1 (Tamura et al., 2011), where dXY is the mean distance between groups X and Y, and dX and dY are mean genetic distance within each group.

| Phylogenetic analysis and population differentiation
The neighbor-joining tree based on the concatenated trnW-L+cox3 sequences revealed substantial phylogeographic structure in S. fusiforme

| Principal component analysis and divergence time
Deep phylogeographic structure in S. fusiforme was also supported by mtDNA-based principal component analysis (PCoA) (Figure 3a). Firstly, we conducted PCoA with all populations using concatenated mtDNA data and found that S. fusiforme populations grouped into three major clusters which were concordant with the groups defined by phylogenetic and network analyses (Figures 1a, 3a). Because there was no clear structure detected within group C (Figures 3a), we then conducted further PCoA analysis for populations (POP 6-26) in this group. As expected, the populations were grouped into three clusters as revealed by phylogenetic analyses (Figures 1a and 3a). Mitochondrial markerbased haplotype network consistently showed genetic divergence between populations from different regions corresponding to three major groups, while the haplotype network and PCoA inferred from RuBisCo rbcL-S spacer showed no phylogeographic structure (Figure 3b, Fig. S3).

| Demographic history
Neutrality tests showed that both Fu's Fs and Tajima's D were negative for groups C2 and C3, indicating a scenario of historical population expansion (Table S5). For groups A, B, and C1, most of the values were statistically nonsignificant. However, the mismatch distribution for each group showed a unimodal except for group B (Fig. S4) Figure 5). The comparison of historical (θ w ) and contemporary (θ π ) genetic diversity showed that θw was substantially higher than θ π , especially for groups C2 and C3 (Table S5), indicating a recent bottleneck.

| Molecular diversity and phylogeographic patterning
In this study, the majority of phylogeographic patterns in S. fusiforme were based on the concatenated mitochondrial DNA rather than the F I G U R E 3 Principal component analysis (PCoA) based on trnW-L+cox3 (a) and haplotype network for each marker (b). Each line between main haplotypes represents one mutation step. The groups marked in PCoA are the same as in Figure 1 T1 T2   T3   T4   T5   T6   T7   T18   T17  T16   T15   T14   T13   T12   T8   T11   T9   T10   T19   T20   T21   T22  plastid rbcL-S. This is principally due to their contrasting phylogeographic resolution (Figure 2, Table S2). Like other oogamous members (e.g., Fucus) in the brown algae, the mitochondria and plastid of Sargassum are inherited maternally (Motomura, Nagasato, & Kimura, 2010). However, rearrangement and duplication events enable mitochondrial genes to have faster mutation rates than plastids (Mattio & Paryi, 2010;Wan, Wu, Fujihara, & Fang, 2004), leading to different capability in detecting phylogenetic diversity and biogeography of Sargassum species (Chan et al., 2013;Cheang et al., 2010;Hu et al., 2011). In general, mitochondrial DNA still provides important insights into biodiversity partitioning and biogeography of S. fusiforme in the Asia-Northwest Pacific (ANP) at a fine spatial-temporal scale.
S. fusiforme is characterized by deep genetic splits in the ANP (Figure 1 and 3), resembling the phylogeographic structure observed in S. horneri (Hu et al., 2011). Such shared structure indicates that both species might have persisted in multiple glacial refugia and phylogeo-  Horiguchi, Hanyuda, & Kawai, 2016). However, the congeneric S. polycystum and S. thunbergii in the ANP exhibit high population homogeneity (Chan et al., 2013;Kantachumpoo, Uwai, Noiraksar, & Komatsu, 2014;Li et al., 2016). These contrasting phylogeographic patterns may reflect distinct evolutionary history occurred in Sargassum species and/or the physical environment it is subjected to. Interestingly, the population of S. fusiforme in the Sea of Japan (JAM) has a mixture of two subgroups (C1 and C2) (Figure 1). It not only harbors the richest diversity, but is also highly divergent from populations in the Japan main islands, rather than populations in China-Korea coasts (Table 1 and Table S2). Biogeographically, the Tsushima Warm Current could bring individuals of S. fusiforme (C2) from the southwestern diversity hotspot (e.g., the Jeju Island) into the Sea of Japan (Li et al., 2016), accounting for the structured diversity. Nevertheless, more sampling in the Sea of Japan is necessary to determine whether it could benefit from transregional migration from adjacent areas driven by currents.
Likewise, the population in southern China (CDS) also exhibits a mixture of the subgroups (C2 and C3). This diversity pattern likely results from unintentional transportation of C2 from the Yellow-Bohai Sea and/or the Sea of Japan mediated by anthropogenic interference, but further survey is needed.

| Conservation insights for S. fusiforme
S. fusiforme at different geographic scales has shown distinct ecological adaptation to local environmental variables (Kokubu et al., 2015;Nagato & Kawaguchi, 2003;Zou et al., 2006): It is therefore important to determine which populations should be used for breeding and selection purposes in the context of marine culture. The phylogeographic heterogeneity in S. fusiforme, with lineages A and B, subgroup C1 distributed around the Japan main islands and subgroups C2 and C3 interspersed along the China coast, respectively (Figure 1), indicates that translocations and long-term monitoring among these genetically isolated units might be a feasible potion to augment locally depleted/extinct natural resource (Moritz, 1999). In reality, natural populations of S. fusiforme in Korea, Japan, and northern China have been collected since 1990s and used as seedlings for farming and cultivation to complement the overexploitation of S. fusiforme in Dongtou County, Zhejiang Province, China (Li et al., 2010). It should be noted, however, that these activities could lead to introductions and hybridization that could be harmful for the fitness of progeny and the survival of local populations (outbreeding depression : Frankham, 2005). In addition, the ecological exchangeability among these geographically isolated populations of S. fusiforme should be examined prior to conservation approach considering their genetic inexchangeability (Crandall et al., 2000;Moritz, 1999;Rader, Belk, Shiozawa, & Crandall, 2005).
The Japan main islands which host a hotspot of genetic diversity can be explained by the maintenance of distinct ancestral relics and relatively long-term demographic stability (Dufresnes et al., 2013).  Table S2). This mode indicates that populations in the Japanese Archipelago are likely to be especially vulnerable to small population effects and fragmentation (Rossiter, Benda, Dietz, Zhang, & Jones, 2007). In particular, the importance of genetic diversity for historically survived relic populations has long been acknowledged (Frankham, 2005;Rabinowitz & Zeller, 2010).
Our results unequivocally show that the south-to-central coasts of the Japan main islands have rich groups and endemic diversity. This area might constitute a few refugia during the Pleistocene ice ages (Hu et al., 2011(Hu et al., , 2015Kamei, 1981;Kojima, Fujikura, & Okutani, 2004a), becoming a region that may have been characterized by processes such as habitat fragmentation, range expansion, and secondary contact (Canestrelli, Climmaruta, Costantini, & Nascetti, 2006). Therefore, this region should be identified as a priority for conservation. On the other hand, clustering-based phylogeny and network analyses (Figure 1 and

| Broadening conservation effort to marine species in the ANP
The ANP, particularly the Japanese Archipelago, is one of the most prominent marine biodiversity hotspots on the earth (Kerswell, 2006;  , 2015;Norton et al., 1996) and should receive specific con-  (Hu et al., 2011;Kamei, 1981;Kojima et al., 2004a) and the variable coastal habitat conditions (Hu et al., 2015). Thus, conservation insights derived from phylogeographic processes of S. fusiforme in this study are also applicable to other marine species inhabiting similar coastal environments around the Japanese Archipelago. Integrating ecological niches, biological features, and life histories of different species as a manageable system may minimize the impact of anthropogenic activities and habitat loss on these diversity hotspots, and ultimately to preserve and maintain coastal ecosystem functioning.

| Defining evolutionary significant units (ESUs)
As a canopy-forming macroalga, S. fusiforme has a fundamental role in structuring the coastal marine community in the ANP. Conserving this single species will contribute significantly to overall biodiversity and community ecosystems. Our study thus raises the important question whether the diverged genetic lineages in S. fusiforme can be defined as conservation units. The concept of ESUs was initially raised in the 1980s by Ryder (1986), and since then, various definitions have been proposed and formulated (Moritz, 1994;Paetkau, 1999). Fraser and Bernatchez (2001)  June to August (Nagato & Kawaguchi, 2003), whereas on the coast of Nanao Island, Guangdong, China, its maturation period continued from mid-April to late June (Zou et al., 2006). Taking these lines of evidence together, the lineages detected in S. fusiforme in this study fulfill the requirements of ESUs.
Nevertheless, there are limitations in applying uniparentally inherited mitochondrial genes for conservation biogeography. First of all, the ESUs identified in S. fusiforme correspond to species or subspecies boundaries in terms of conservation genetics and taxonomic uncertainties (Karl & Bowen, 1999), and we should be cautious with gene-specific, lineage-specific, and population-specific evolution in mitochondria (Wan et al., 2004). Moreover, Moritz (1994)

| CONCLUDING REMARKS
The structured genetic lineages and sublineages with nearly disjunct geographic boundaries found in S. fusiforme reflect both older phylogeographic history and long-term isolation. These historical phylogeography and cryptic genetic diversity can not only help to understand species' underlying adaptive potentials following drastic environmental shifts, but also present important insights for conservation purpose and predict how species will respond to future climate change.
Nevertheless, ancestral genetic relics have been demonstrated to link to population trends in a biogeographic context (Schmitt & Hewitt, 2004). Geographically, peripheral and central populations have shown variable genetic features and adaptive adjustments to local environments in the coastal marine community (Diekmann & Serräo, 2012;Viejo, Marinez, Arrontes, Astudillo, & Hernández, 2011;Zardi et al., 2015). To maximize conservation effectiveness, research toward linking these evolutionary processes and landscape change to phenotypic performance and fitness emerges as an urgent need to rank lineages of S. fusiforme on a scale from less to more susceptible to damage and to recognize protection priorities.