| |

Received: 13 September 2017    Revised: 26 July 2018    Accepted: 13 August 2018

DOI: 10.1111/eva.12696

ORIGINAL ARTICLE

Population genetic structure of the deep-­sea mussel

Bathymodiolus platifrons (Bivalvia: Mytilidae) in the Northwest
Pacific

Ting Xu1 | Jin Sun2 | Hiromi K. Watanabe3 | Chong Chen3  | Masako Nakamura4 | 

Rubao Ji5 | Dong Feng6 | Jia Lv7 | Shi Wang7,8 | Zhenmin Bao7,9 | 
Pei-Yuan Qian2 | Jian-Wen Qiu1

1
Department of Biology, Hong Kong Baptist
University, Hong Kong, China Abstract
2
Department of Ocean Science, Hong Kong Studying population genetics of deep-­sea animals helps us understand their history
University of Science and Technology, Hong
of habitat colonization and population divergence. Here, we report a population ge-
Kong, China
3 netic study of the deep-­sea mussel Bathymodiolus platifrons (Bivalvia: Mytilidae)
Japan Agency for Marine-Earth Science and
Technology (JAMSTEC), Yokosuka, Japan widely distributed in chemosynthesis-­based ecosystems in the Northwest Pacific.
4
School of Marine Science and Technology, Three mitochondrial genes (i.e., atp6, cox1, and nad4) and 6,398 genomewide single
Tokai University, Shizuoka, Japan
5
nucleotide polymorphisms (SNPs) were obtained from 110 individuals from four hy-
Department of Biology, Woods Hole
Oceanographic Institution, Woods Hole, drothermal vents and two methane seeps. When using the three mitochondrial
Massachusetts genes, nearly no genetic differentiation was detected for B. platifrons in the Northwest
6
CAS Key Laboratory of Ocean and Marginal
Pacific. Nevertheless, when using SNP datasets, all individuals in the South China Sea
Sea Geology, South China Sea Institute of
Oceanology, Chinese Academy of Sciences, (SCS) and three individuals in Sagami Bay (SB) together formed one genetic cluster
Guangzhou, China
that was distinct from the remaining individuals. Such genetic divergence indicated a
7
Ministry of Education Key Laboratory of
Marine Genetics and Breeding, College of
genetic barrier to gene flow between the SCS and the open Northwest Pacific, re-
Marine Life Sciences, Ocean University of sulting in the co-­occurrence of two cryptic semi-­isolated lineages. When using 125
China, Qingdao, China
8
outlier SNPs identified focusing on individuals in the Okinawa Trough (OT) and SB, a
Laboratory for Marine Biology and
Biotechnology, Qingdao National Laboratory minor genetic subdivision was detected between individuals in the southern OT (S-­
for Marine Science and Technology, OT) and those in the middle OT (M-­OT) and SB. This result indicated that, although
Qingdao, China
9 under the influence of the Kuroshio Current and the North Pacific Intermediate
Laboratory for Marine Fisheries Science
and Food Production Processes, Qingdao Water, subtle geographic barriers may exist between the S-­OT and the M-­OT.
National Laboratory for Marine Science and
Introgression analyses based on these outlier SNPs revealed that Hatoma Knoll in the
Technology, Qingdao, China
S-­OT represents a possible contact zone for individuals in the OT-­SB region.
Correspondence
Furthermore, migration dynamic analyses uncovered stronger gene flow from Dai-­
Jian-Wen Qiu, Department of Biology, Hong
Kong Baptist University, Hong Kong, China. yon Yonaguni Knoll in the S-­OT to the other local populations, compared to the re-
Email: [email protected]
verse directions. Taken together, the present study offered novel perspectives on the
Funding information genetic connectivity of B. platifrons mussels, revealing the potential interaction of
This project was supported by a General
Research Fund from University Grants ocean currents and geographic barriers with adaption and reproductive isolation in
Committee (12302917) and a Strategic shaping their migration patterns and genetic differentiation in the Northwest Pacific.
Development Fund from Hong Kong Baptist
University (15-1012-P04) to JWQ, a HKBU
PhD studentship to TX, and an InterRidge
Fellowship to JS.

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium,
provided the original work is properly cited.
© 2018 The Authors. Evolutionary Applications published by John Wiley & Sons Ltd

Evolutionary Applications. 2018;11:1915–1930. wileyonlinelibrary.com/journal/eva |  1915

  
 |
1916       XU et al.

KEYWORDS
Bathymodiolus, deep-sea, genetic structure, introgression, migration patterns, mitochondrial
genes, population connectivity, RAD-seq

1 |  I NTRO D U C TI O N Most marine benthic animals, including those in the deep ocean,
have a biphasic life with a pelagic larval stage through which they
Hydrothermal vents and cold seeps generally occur in tectonically achieve connectivity across different habitats (Cowen & Sponaugle,
active areas and along continental margins, where neighboring 2009). Knowledge on population connectivity of vent and seep an-
sites are often separated by tens to hundreds of kilometers in the imals sheds light on the scale, direction, and frequency of dispersal,
ocean (Le Bris et al., 2016). Despite differences in water tempera- which will not only enhance our understanding of the mechanisms
ture and main source of fluid, vent and seep ecosystems are both shaping their global and regional biogeography, but also provide key
fueled mainly by chemosynthesis, the conversion of carbon dioxide insights into their recovery potential in response to environmental and
and/or methane into organic matters in microbes via oxidation of anthropogenic disturbances (Baco et al., 2016; Kinlan & Gaines, 2003;
reduced substances, such as hydrogen sulfide, methane, and hydro- Miller, Thompson, Johnston, & Santillo, 2018; Rogers et al., 2012).
gen, unlike shallow-­water ecosystems that are driven primarily by Deep-­sea mussels in the genus Bathymodiolus (Bivalvia:
photosynthesis (Tunnicliffe, Juniper, & Sibuet, 2003). High chemo- Mytilidae) are one of the most iconic, dominant, and important
synthetic primary production enables these ecosystems to support a foundation taxa in chemosynthesis-­based ecosystems (Van Dover,
much higher biomass and abundance of megafauna compared to the 2000). Dense Bathymodiolus mussel beds generate a highly complex
surrounding seabed (Levin et al., 2016). habitat for a variety of other animals to inhabit (Figure 1a; Bruno

F I G U R E   1   Distribution of B. platifrons in the Northwest Pacific. (a) A representative photograph of JR showing the B. platifrons and
Shinkaia crosnieri (squat lobster) dominated community. (b) Sampling locations of B. platifrons and the dominant ocean currents in the study
region. Vents and seeps are represented in red and green dots, respectively. Ocean currents were redrawn based on You et al. (2005) and
Nan et al. (2011). Three flow patterns of the Kuroshio Current when passing through the Luzon Strait, namely the leaping, the leaking, and
the looping path, are indicated by 1, 2, and 3, respectively. Code: DK, Dai-­yon Yonaguni Knoll; ECS, East China Sea; HK, Hatoma Knoll; IN,
Iheya North; IR, Iheya Ridge; JR, Jiaolong Ridge; KC, Kuroshio Current; NPIW, North Pacific Intermediate Water; NWP, Northwest Pacific;
OH, Off Hatsushima; SCS, South China Sea
XU et al. |
      1917

& Bertness, 2001; Govenar, 2010; Vrijenhoek, 2010). Bathymodiolus seeps included Jiaolong Ridge (JR; 1,122 m depth) in the SCS and
mussels produce planktotrophic larvae capable of migrating to Off Hatsushima (OH; two sites in 858 m and 1,172 m depth) in
surface water and dispersing across a long distance in ocean cur- Sagami Bay (SB). These sampling locations span a horizontal distance
rents with very long planktonic larval durations (Arellano, Van of more than 2,400 km. Upon arrival on the deck, mussels were ei-
Gaest, Johnson, Vrijenhoek, & Young, 2014; McVeigh, Eggleston, ther dissected immediately for preservation in 95%–100% ethanol
Todd, Young, & He, 2017; Young et al., 2012). To date, 30 species or frozen immediately at −80°C for later dissection. Genomic DNA
including eight fossil species of Bathymodiolus have been reported was extracted from the adductor muscle of each individual using
(MolluscaBase, 2018), although the genus appears to be polyphyletic the phenol/chloroform extraction protocol (Sambrook, Fritsch, &
(Lorion et al., 2014). Among them, mussels of the Northwest Pacific Maniatis, 1989). Concentration and purity of the extracted DNA
species Bathymodiolus platifrons are considered to be a good candi- were measured using a NanoDrop ND-­1000 spectrophotometer
date for population genetic studies due to their wide horizontal (22° (Thermo Fisher Scientific, Wilmington, DE, USA), and the integrity
to 35°N) and bathymetric (642 to 1,684 m) distribution ranges, as of DNA was checked by 1.0% agarose gel electrophoresis.
well as their capability to inhabit both hydrothermal vents and meth-
ane seeps (Fujikura et al., 2007; Suess, 2005; Watanabe, Fujikura,
2.2 | Mitochondrial gene amplification and genetic
Kojima, Miyazaki, & Fujiwara, 2010; http://www.godac.jamstec.
statistic estimation
go.jp/bio-sample/index_e.html, April 2018).
Previous studies based on one or several mitochondrial genes re- The newly designed primer pair BP_atp6F (5′-­C ATAGG
vealed a lack of genetic differentiation among vent and seep popula- AGCAAAGTAAGTGG-­3′) and BP_atp6R (5′-­GGTTCTACCA
tions of B. platifrons (Kyuno et al., 2009; Miyazaki et al., 2013; Shen CCATCCTCG-­3′), the universal primer pair LCO1490 and HCO2198
et al., 2016). Nevertheless, our recent study based on 9,307 genome- (Folmer, Black, Hoeh, Lutz, & Vrijenhoek, 1994), and the primer pair
wide single nucleotide polymorphisms (SNPs) generated by the type ArgBL and NAP2H (Arèvalo, Davis, & Sites, 1994; Bielawski & Gold,
IIB restriction site-­associated DNA (2b-­R AD) approach detected a 1996) were used to amplify sequences of atp6, cox1, and nad4 genes,
clear genetic divergence between individuals of B. platifrons from a respectively. Polymerase chain reactions (PCRs) were performed
vent field in the Okinawa Trough (OT) and a methane seep in the South using a Mastercycler Gradient thermocycler (Eppendorf, Germany)
China Sea (SCS) (Xu et al., 2017). However, due to the difference in with the following program: 2 min initial denaturation at 94°C, 30
results based on different genetic marker types and limited sampling cycles of 30 s denaturation at 94°C, 30 s annealing at 50°C, 30 s
locations in these studies, the population structure and migration pat- extension at 72°C, and a final extension for 10 min at 72°C. The PCR
terns of B. platifrons in the Northwest Pacific remain puzzling. products were separated using 1.0% agarose gel electrophoresis
Therefore, we carried out a population genetic study of B. plat- and purified using the Zymoclean™ Gel DNA Recovery Kit (Zymo
ifrons combining both mitochondrial genes and genomewide SNP Research, Irvine, CA, USA) following the manufacturer’s protocol.
markers derived from samples from all representative habitats of Purified PCR products were bidirectionally sequenced on an ABI
this species known thus far. Through this study, we aimed to better PRISM® 3730xl DNA Analyzer. Sequences obtained were visually
understand the patterns and causes of genetic connectivity, genetic checked and assembled into contigs using the DNASTAR Lasergene
divergence, and migration dynamics of B. platifrons, the most widely package (DNASTAR Inc., Madison, WI, USA). Each mitochondrial
distributed Bathymodiolus mussel in the Northwest Pacific. gene from all individuals was aligned using MUSCLE v.3.8.31 (Edgar,
2004) under the default settings and manually trimmed to the same
length for subsequent analyses.
2 | M ATE R I A L S A N D M E TH O DS Pairwise sequence divergence was calculated based on the
Kimura-­2-­parameter (K2P) model (Kimura, 1980) implemented in
MEGA v.7 (Kumar, Stecher, & Tamura, 2016) for each mitochondrial
2.1 | Sample collection and DNA extraction
gene. For individuals collected from each location (i.e., local popu-
A total of 110 adults of B. platifrons used in this study were col- lation), the number of haplotypes (H), haplotype diversity (Hd), and
lected from four hydrothermal vents and two methane seeps be- nucleotide diversity (π) of each mitochondrial gene were estimated
tween 2009 and 2014 (Figure 1b), either by the remotely operated using Arlequin v.3.5.2.2 (Excoffier & Lischer, 2010). Two neutrality
vehicle (ROV) Hyper-Dolphin on-­board the research vessels (R/Vs) tests, including Tajima’s D (Tajima, 1989) and Fu’s FS (Fu, 1997), im-
Natsushima and Kaiyo of Japan Agency for Marine-­Earth Science plemented in Arlequin, were conducted on each mitochondrial gene
and Technology (JAMSTEC), or by the manned deep-­submergence of each local population, with 10,000 simulations performed to test
vehicle Jiaolong on-­board the Chinese R/V Xiangyanghong 9 (see for significance.
Supporting Information Table S1 for details). The four hydrothermal In addition, the three mitochondrial genes of each individual
vents included Dai-­yon Yonaguni Knoll (DK; 1,344 m depth) and were concatenated into a single sequence using SequenceMatrix
Hatoma Knoll (HK; 1,482 m depth) in the southern OT (S-­OT), Iheya v.1.7.8 (Vaidya, Lohman, & Meier, 2011), and the concatenated se-
Ridge (IR; 1,402 m depth) and Iheya North (IN; two sites in 993 m quences were then used to estimate pairwise FST between local pop-
and 1,002 m depth) in the middle OT (M-­OT); the two methane ulations using Arlequin, with 10,000 permutations applied to test for
 |
1918       XU et al.

significance. Furthermore, to investigate the population structure, by the exact test implemented in Arlequin for each local population,
TCS haplotype networks were reconstructed for each mitochondrial with 100,000 dememorization steps followed by 1,000,000 steps in
gene using POPART v.1.7 (Leigh & Bryant, 2015). a Markov chain; (h) number of SNPs per locus ≤3.
Genomewide genetic statistics, including expected heterozygos-
ity (Hexp), Hobs, π, and inbreeding coefficient (FIS), were calculate using
2.3 | RAD library construction, sequencing, and
Stacks. Two neutrality tests, including Tajima’s D (Tajima, 1989) and
data filtering
Fu and Li’s D* (Fu & Li, 1993), for each polymorphic loci of each local
The 2b-­R AD method (Wang, Meyer, McKay, & Matz, 2012) was ap- population, were estimated based on the batch mode implemented
plied to construct RAD libraries. Genomic DNA from each individ- in DnaSP v.5 (Rozas, Sánchez-­DelBarrio, Messeguer, & Rozas, 2003).
ual was digested using the type IIB restriction enzyme BsaXI (New Filtered SNP datasets were further formatted using the
England BioLabs, Ipswich, MA, USA). Two adaptors with compatible POPULATIONS module in Stacks, FORMATOMATIC v.0.8.1
(5′-­NNN-­3′) overhangs were used to link the digested products, and (Manoukis, 2007), and/or PGDSpider v.2.0.8.3 (Lischer & Excoffier,
a 6-­bp unique barcode was then added to each individual, result- 2012) for downstream analyses.
ing in DNA libraries of approximately 155 bp. These libraries were
afterward purified using the QIAquick PCR Purification Kit (Qiagen,
2.5 | Outlier SNP detection and characterization
Chatsworth, CA, USA), analyzed for integrity by 8.0% polyacryla-
mide gel electrophoresis (PAGE), quantified using the Qubit 2.0 The coalescent method implemented in Arlequin was used to screen
Fluorometer (Invitrogen, Carlsbad, CA, USA), and then pooled for for candidate outlier SNPs. It has been reported that the hierarchical
single-­end sequencing on the Illumina HiSeq 1500, HiSeq 2000, island model in this software is more powerful than the finite island
and/or HiSeq 4000 platforms. model in outlier SNP detection for hierarchically subdivided popu-
Raw sequencing reads were filtered using a custom Perl script to lations or populations with a recent common ancestry (Excoffier,
remove adaptors and the 3-­bp terminal sequences of each read (Jiao Hofer, & Foll, 2009). Therefore, based on the geographic affinity and
et al., 2014). Reads with ≥10 nucleotide positions having a Phred the result of principal component analyses (PCA) carried out using
quality index <20, without restriction sites, with ambiguous bases, the entire SNP dataset (see section 3.4), two hierarchical island mod-
or ≥30% homopolymer regions, were all discarded. els were assumed in Arlequin to screen for candidate outlier SNPs:
(a) individuals were divided into two groups: SCS = JR, and the OT-­SB
region = DK, HK, IR, IN, and OH; (b) individuals in JR and the three
2.4 | SNP identification and genetic
JR-­like individuals in the OH site of SB were excluded, and the re-
statistic estimation
maining individuals were divided into three groups: S-­OT = DK and
Based on the recognition sites of the BsaXI enzyme, 2b-­R AD tags HK, M-­OT = IR and IN, and SB = OH. The outlier screening analyses
were extracted from the draft genome of B. platifrons (Sun et al., were then carried out by running 100,000 simulations, along with
2017), which served as a reference for SNP identification. Alignment 100 simulated demes and 50 stimulated groups.
of the filtered reads of each individual to the reference was achieved Genomic regions of the identified candidate outlier SNPs were
by SOAP v.2.21 (Li et al., 2009) using the match mode of “find the determined by mapping the 2b-­R AD tags that harbored outlier SNPs
best hits” (-­M 4), the maximum number of allowed mismatches of against the draft genome of B. platifrons (Sun et al., 2017). For those
two (-­v 2), and no repeat allowed (-­r 0). The output file for each indi- mapped to the genic regions [i.e., coding DNA sequence (CDS), in-
vidual was converted into the.sam format using soap2sam.pl (http:// tron, or 3/5′-­untranslated region (3/5′-­UTR)], their corresponding
soap.genomics.org.cn/soapaligner.html). proteins (i.e., outlier-­associated proteins) and annotations (Sun et al.,
Detection of SNPs was carried out using the ref_map.pl pipe- 2017) were extracted for functional classification.
line implemented in Stacks v.1.41 (Catchen, Hohenlohe, Bassham,
Amores, & Cresko, 2013) with the following criteria: (a) ≥10 aligned
2.6 | Genetic differentiation estimation and its
reads to build a stack (-­m 10); (b) loci be biallelic as those with more
relatedness to geographic distance
alleles are likely caused by sequencing or clustering errors; (c) loci
with a depth coverage (i.e., number of reads of a specific locus that Values of pairwise FST between local populations were estimated
matched to the reference) ≤120 to reduce bias derived from repet- using Arlequin based on the entire SNP dataset and the two outlier
itive genomic contents; (d) loci present in all local populations and SNP datasets, with 10,000 permutations to determine significance.
genotyped for ≥70% individuals in each local population; (e) SNPs The Mantel test implemented in the same software was carried
with an overall minor allele frequency (MAF) ≥0.02 to reduce PCR out to correlate genetic distance (i.e., values of pairwise FST cal-
and sequencing errors, as well as uninformative markers (Roesti, culated based on the entire SNP dataset) and geographic distance
Salzburger, & Berner, 2012); (f) loci with an observed heterozygos- (km), also with 10,000 permutations applied to test for significance.
ity (Hobs) ≤0.5 among all individuals to avoid inclusion of paralogs The approximate geographic distance between each pair of local
(Hohenlohe, Amish, Catchen, Allendorf, & Luikart, 2011); (g) SNPs populations was measured using the Latitude/Longitude Distance
conforming to Hardy–Weinberg equilibrium (p ≥ 0.01) as assessed Calculator (http://jan.ucc.nau.edu/~cvm/latlongdist.html). An
XU et al. |
      1919

intermediate geographic point between the two sampling sites of IN population 1, while those in the M-­OT and SB (the three JR-­like in-
and OH was separately generated to simplify the calculation. dividuals in SB were excluded) were together designated as parental
population 2. Values of h refer to the proportion of the genome of
a given individual in the local population of HK that was inherited
2.7 | Population structure and individual assignment
from the designated parental population 2 (Buerkle, 2005; Gompert
based on the entire SNP dataset and the two outlier
& Alex Buerkle, 2010).
SNP datasets
A Bayesian approach implemented in STRUCTURE v.2.3.4 (Pritchard,
2.9 | Migration dynamic analyses based on the
Stephens, & Donnelly, 2000) was applied to detect population struc-
entire SNP dataset
ture. The LOCPRIOR model which uses sampling locations as prior
information to assist the clustering and the corrected allele frequen- The web-­based software divMigrate-­online (Sundqvist, Keenan,
cies model were selected. The number of genetic clusters K was set Zackrisson, Prodöhl, & Kleinhans, 2016) was applied to infer the di-
from 1 to 6, each with five replicates, using a burn-­in of 100,000 rectional relative migration patterns using the GST statistic (Nei, 1973)
followed by 1,000,000 iterations. The optimal K was evaluated using as a measure of genetic differentiation. The method implemented in
STRUCTURE HARVESTER v.0.6.94 (Earl & vonHoldt, 2012). An opti- this software is based on defining a hypothetical pool of migrants for
mal alignment of replicate runs at the optimal K was determined using a given pair of populations and estimating an appropriate measure
CLUMPP v.1.1.2 (Jakobsson & Rosenberg, 2007), and the graph of of genetic differentiation between each of the two populations and
genetic structure was visualized using DISTRUCT v.1.1 (Rosenberg, the hypothetical pool (Sundqvist et al., 2016). The directional genetic
2004). The PCA implemented in the R package SNPRelate (Zheng differentiation can be used afterward to evaluate the relative levels
et al., 2012) were used to perform individual assignment based on of migration between the two populations. The larger of the two rela-
the genetic variation among individuals. tive migration values indicates the population is most likely the source
All of the above analyses were carried out using both the entire population, whereas the smaller of the two values indicates the popu-
SNP dataset and the two outlier SNP datasets separately, with only lation is most likely to be the sink population (Sundqvist et al., 2016).
one SNP per locus retained in each dataset to avoid bias derived
from potential linkage disequilibrium.
3 | R E S U LT S

2.8 | Introgression analyses based on the outlier 3.1 | Population genetic analyses based on

SNP dataset mitochondrial genes
Since two genetic backgrounds were detected in the local population Alignment and trimming of the amplified sequences resulted in 717-­bp
of HK based on the second outlier SNP dataset (see section 3.7), the full-­length atp6, 647-­bp partial cox1, and 597-­bp partial nad4 gene se-
USEPOPINFO model which uses sampling locations to screen for mi- quences. Pairwise sequence divergence ranged from 0 to 0.84% (mean:
grants or hybrids in STRUCTURE was applied to test the hypothesis of 0.20%) for atp6 (Supporting Information Table S2), 0 to 0.78% (mean:
mixed ancestry for individuals in HK during the past two generations 0.17%) for cox1 (Supporting Information Table S3), and 0 to 1.02%
(i.e., GENSBACK = 2). Two clusters (i.e., K = 2) were defined according (mean: 0.22%) for nad4 (Supporting Information Table S4). Values of
to the results of STRUCTURE and PCA based on the second outlier Hd for each local population ranged from 0.7273 to 0.9778 for atp6,
SNP dataset (see section 3.7), with one containing individuals in the S-­ 0.4909 to 0.8286 for cox1, and 0.6476 to 0.8667 for nad4; values of π
OT, and the other containing those in the M-­OT and SB with the three ranged from 0.0015 to 0.0030 for atp6, 0.0008 to 0.0022 for cox1, and
JR-­like individuals in SB excluded. The program was run with a burn- 0.0015 to 0.0027 for nad4 (Table 1). Most Tajima’s D and all Fu’s FS sta-
­in of 100,000 followed by 1,000,000 iterations based on the second tistics were significantly (p < 0.05) negative for local populations of B.
outlier SNP dataset (only one SNP per locus retained). Three different platifrons (Table 1). Values of pairwise FST calculated based on the three
values of MIGRPRIOR, which is v, were applied to test whether the concatenated mitochondrial genes ranged from −0.0134 to 0.0782,
results were robust (Pritchard et al., 2000). Individuals with less than with no statistical significance detected after Bonferroni correction
50% posterior probability of having pure ancestry from the designated (Table 2). TCS haplotype networks based on each mitochondrial gene
population (i.e., the S-­OT) were considered to be migrants or hybrid de- all roughly exhibited a star-­like shape, with the most frequent haplo-
scendants (Falush, Stephens, & Pritchard, 2007; Pritchard et al., 2000). type shared among different locations in the center being surrounded
Introgression signature in the local population of HK was fur- by several low frequency and private haplotypes (Figure 2).
ther investigated by calculating the hybrid index (h) (Buerkle, 2005)
using the R package INTROGRESS (Gompert & Alex Buerkle, 2010),
3.2 | SNP identification and genetic
also based on the second outlier SNP dataset (only one SNP per
statistic estimation
locus retained). According to the population structure revealed by
STRUCTURE analyses based on the second outlier SNP dataset (see Sequencing of 2b-­R AD libraries generated approximately 2.2 bil-
section 3.7), the local population of DK was designated as parental lion reads in total, with a mean of 19.9 million reads per individual.
 |
1920       XU et al.

TA B L E   1   Summary genetic statistics of

Gene Location N H Hd π Tajima’s D Fu’s FS
each local population based on three
atp6 (717 bp) JR 30 12 0.7356 0.0016 −2.1248** −9.2918*** mitochondrial genes
**
DK 20 10 0.7579 0.0016 −2.0474 −7.6,398***
HK 15 8 0.8286 0.0020 −1.5158 −4.3642**
*
IR 10 9 0.9778 0.0030 −1.7295 −7.0432***
IN 11 6 0.7273 0.0015 −1.8506* −3.3039**
**
OH 24 13 0.8478 0.0023 −2.0706 −9.6958***
cox1 (647 bp) JR 30 12 0.7655 0.0022 −1.8726* −7.8294***
**
DK 20 10 0.7105 0.0014 −2.2189 −9.3321***
HK 15 8 0.8286 0.0020 −1.4664 −4.9120***
IR 10 4 0.5333 0.0009 −1.5622* −1.9637**
*
IN 11 4 0.4909 0.0008 −1.6000 −2.0423**
OH 24 10 0.7065 0.0016 −1.9029* −7.7136***
*
nad4 (597 bp) JR 30 12 0.8184 0.0021 −1.7635 −8.7518***
DK 20 13 0.8526 0.0025 −2.1633** −12.1714***
HK 15 6 0.6476 0.0015 −1.4512 −3.2353**
IR 10 7 0.8667 0.0023 −1.8391** −4.5227***
IN 11 6 0.8000 0.0018 −1.4646 −3.4118**
OH 24 15 0.8659 0.0027 −2.0000** −14.2081***

Notes. π: nucleotide diversity; H: number of haplotypes; Hd: haplotype diversity; N: sample size.
Significance: *p < 0.05; **p < 0.01; ***p < 0.001.

TA B L E   2   Pairwise FST estimated based

JR DK HK IR IN OH
on the three concatenated mitochondrial
JR — 0.0038 0.0052 0.0008 0.0103 −0.0097 genes (above diagonal) and the entire set
DK 0.0201 *
— 0.0255 −0.0124 −0.0087 −0.0041 of 6,398 SNPs (below diagonal)

HK 0.0188* 0.0013 — 0.0048 0.0782 −0.0014

*
IR 0.0162 0.0032 −0.0008 — 0.0138 −0.0134
IN 0.0206* 0.0049 −0.0008 0.0004 — 0.0008
*
OH 0.0139 0.0022 −0.0010 0.0008 −0.0005 —

Note. Significance: *p < 0.00001 after Bonferroni correction.

Quality control reduced the data to a mean of 15.5 million reads observed in each local population after filtering: 19 in JR, one in HK,
per individual, providing an average depth coverage of 46.3 × three in OH, and none in DK, IR, or IN. Moreover, local populations
(Supporting Information Table S5). A total of 314,151 2b-­R AD tags of DK and HK in the S-­OT had lower values of Hobs and π as well as
were extracted from the 1.64 Gb genome (Sun et al., 2017), with higher values of FIS compared to those in the SCS, the M-­OT, and SB.
one cutting site in every 5.2 kb. Among these tags, 177,365 (56.5%) All these statistics were summarized in Table 4.
were unique, 965 (0.3%) contained ambiguous bases (N), and 28,161 No polymorphic loci were detected to have a significantly nega-
(43.2%) had repeats (repeat range: 2 to 1,378). All of them were used tive value of Tajima’s D in any local populations, whereas a very small
as the reference for subsequent SNP identification. proportion of polymorphic loci (i.e., <0.4%) were detected to have a
After genotyping and strict filtering, a total of 6,398 SNPs were significantly (p < 0.05) positive value of Tajima’s D in each local pop-
identified from 5,458 2b-­R AD tags (Table 3; Supporting Information ulation (Supporting Information Table S7). Besides, less than 1.1%
Table S6). The number of polymorphic nucleotide sites for the six of polymorphic loci were detected to have a significantly (p < 0.05)
local populations varied from 4,452 (accounting for 3.0% of the negative value of Fu and Li’s D* in the local populations of JR, DK,
total nucleotide sites; IR) to 5,769 (accounting for 3.9% of the total HK, and OH; nevertheless, no polymorphic loci were detected to
nucleotide sites; JR). Since loci were retained only when they were have a value of Fu and Li’s D* with statistical significance in the local
detected in all local populations, only a few private SNPs were population of IR or IN (Supporting Information Table S8).
XU et al. |
      1921

atp6 TA B L E   3   Number of the putative loci retained following each

step of filtering and the final result of candidate SNPs

Category Count

Putative loci after filtering steps

Putative 2b-­R AD loci (stacks depth ≥10) 159,223
Polymorphic loci 91,487
Biallelic loci 76,317
Coverage depth ≤120 54,554
Present in all six populations and genotyped 12,981
individuals in each population ≥70%
Overall minor allele frequency (MAF) ≥0.02 5,815
Observed heterozygosity (Hobs) ≤0.5 5,661
Hardy–Weinberg equilibrium (p ≥ 0.01) 5,464
cox1
Number of SNPs per locus ≤3 5,458
Entire candidate SNPs
Total number 6,398
Number when one per locus was retained 5,458
Candidate outlier SNPs identified by Arlequin (p < 0.01)
SCS and OT-­SB
Total number 106
Number when one per locus was retained 99
S-­OT, M-­OT, and SB
Total number 138
Number when one per locus was retained 125

nad4

all the others in the OT-­SB region (FST range: 0.0139 to 0.0206) after

10 samples
Bonferroni correction (Table 2).
The Mantel test revealed no correlation between genetic
1 sample
distance represented by values of pairwise FST calculated based
on the entire set of 6,398 SNPs and the geographic distance
(Supporting Information Figure S1), showing no evidence for
isolation-­by-­d istance throughout the known distribution range
of B. platifrons.

3.4 | Population structure and individual assignment

F I G U R E   2   TCS haplotype networks inferred based on atp6,
cox1, and nad4. The number of hatch marks along edges indicates
based on the entire SNP dataset
the number of nucleotide substitutions. Color of each circle
STRUCTURE analyses based on the entire set of 5,458 SNPs (only
represents the sampling location where the haplotype was found,
one SNP per locus retained) revealed the occurrence of two ge-
and size of each circle is proportional to the frequency of the
respective haplotype. Black circles indicate unknown or missing netic groups (i.e., optimal K = 2) of B. platifrons in the Northwest
haplotypes Pacific (Supporting Information Figure S2a,b). One genetic group
consisted of all individuals in JR of the SCS as well as three indi-
viduals in OH (i.e., OH1_4, OH1_7, and OH1_9) of SB, and the other
3.3 | Genetic differentiation calculated based
genetic group was formed by the remaining individuals in the OT-­
on the entire SNP dataset and its relatedness to
SB region (Figure 3a). This pattern of population genetic structure
geographic distance
was also detected in the result of PCA along the first eigenvector
Values of pairwise FST calculated based on the entire set of 6,398 (Figure 3b). In addition, there was a minor genetic subdivision be-
SNPs ranged from −0.0010 to 0.0206, with statistical significance tween most individuals in the S-­OT and those in the M-­OT and SB
(p < 0.00001) detected only between the local population of JR and along the second eigenvector (Figure 3b).
1922       | XU et al.

TA B L E   4   Summary genetic statistics of each local population based on the entire set of 6,398 SNPs

Variant positions

Region Location Private Variant sites Poly sites Hexp Hobs π FIS

SCS JR 19 6,398 5,769 0.1662 0.1620 0.1694 0.0292

S-­OT DK 0 6,398 5,227 0.1553 0.1480 0.1594 0.0447
HK 1 6,398 4,914 0.1586 0.1516 0.1648 0.0459
M-­OT IR 0 6,398 4,452 0.1611 0.1635 0.1702 0.0193
IN 0 6,398 4,656 0.1631 0.1631 0.1710 0.0246
SB OH 3 6,398 5,683 0.1664 0.1633 0.1703 0.0278

All (variant and fixed) positions

Region Location Private Total sites % Poly Hexp Hobs π FIS

SCS JR 19 147,366 3.9 0.0072 0.0070 0.0074 0.0013

S-­OT DK 0 147,366 3.5 0.0067 0.0064 0.0069 0.0019
HK 1 147,366 3.3 0.0069 0.0066 0.0072 0.0020
M-­OT IR 0 147,366 3.0 0.0070 0.0071 0.0074 0.0008
IN 0 147,366 3.2 0.0071 0.0071 0.0074 0.0011
SB OH 3 147,366 3.9 0.0072 0.0071 0.0074 0.0012

Notes. % Poly: percentage of polymorphic sites in total nucleotide sites (i.e., total sites); FIS: inbreeding coefficient; Hexp: expected heterozygosity; Hobs:
observed heterozygosity; Poly sites: number of polymorphic sites; Private: number of unique SNPs; π: nucleotide diversity.
Total sites: 27 nucleotide sites per locus × 5,458 loci = 147,366 nucleotide sites.

3.5 | Identified outlier SNPs based on two

(a)
hierarchical island models
By defining a hierarchical island model composed of the SCS and the
OT-­SB region in Arlequin, a total of 106 candidate outlier SNPs asso-
K=2

ciated with 99 loci (p < 0.01) were identified (Figure 4a), with the al-
lele frequency of each outlier SNP shown in Supporting Information
Table S9. Among them, 32 outlier SNPs associated with 30 loci were
JR DK HK IR IN OH
mapped to 27 proteins derived from the draft genome of B. platifrons
(b) 0.4 (Sun et al., 2017). These outliers were found in different regions
JR
DK throughout the genome, including six (5.7%) in CDSs with two being
HK
IR
IN
nonsynonymous substitutions, 20 (18.9%) in introns, and six (5.7%)
Eigenvector 2 (1.3%)

OH
0.2 in 3′-­UTRs. The outlier-­associated proteins were manually classified
into ten broad categories, including biological adhesion (2, 7.4%), car-
bohydrate and lipid metabolism (2, 7.4%), cell death (1, 3.7%), DNA
0.0 metabolism (4, 14.8%), localization (4, 14.8%), peptide metabolism
(1, 3.7%), response to stimulus (3, 11.1%), signaling (3, 11.1%), sys-
tem development and processing (2, 7.4%), and those with unknown
–0.2 functions (5, 18.6%) (Supporting Information Table S10).
By defining a hierarchical island model composed of the S-­OT,
–0.2 –0.1 0.0 0.1 0.2 0.3 the M-­OT, and SB, a total of 138 candidate outlier SNPs associated
Eigenvector 1 (2.0%) with 125 loci (p < 0.01) were identified (Figure 4b), with the allele
frequency of each outlier SNP shown in Supporting Information
F I G U R E   3   Population genetic structure of B. platifrons inferred
based on the entire set of 5,458 SNPs (only one SNP per locus Table S11. A total of 46 outlier SNPs associated with 39 loci were
retained) using (a) STRUCTURE analyses and (b) PCA implemented in mapped to 38 proteins derived from the draft genome of B. plati-
SNPRelate. In (a), each individual is represented by a single bar, with frons (Sun et al., 2017). Among these outliers, 11 (8.0%) were found
different colors showing membership fractions of each inferred cluster in CDSs with five being nonsynonymous substitutions, 28 (20.3%)
XU et al. |
      1923

(a) 0.6 (b) 0.3

0.5

0.4 0.2

0.3
FST

FST
0.1
0.2

0.1
0.0
0.0

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.0 0.1 0.2 0.3 0.4 0.5
Heterozygosity Heterozygosity

(c) (d)
K=2

K=2
JR DK HK IR IN OH DK HK IR IN OH

(e) (f) 0.6

JR DK
0.4 DK HK
HK IR
IR 0.4 IN
IN OH
Eigenvector 2 (4.1%)

Eigenvector 2 (5.7%)

OH
0.2
0.2

0.0 0.0

–0.2
–0.2

–0.4
–0.4
–0.6
–0.2 –0.1 0.0 0.1 0.2 0.3 –0.3 –0.2 –0.1 0.0 0.1 0.2 0.3 0.4
Eigenvector 1 (22.1%) Eigenvector 1 (8.7%)

F I G U R E   4   Population genetic structure of B. platifrons inferred based on the two outlier SNP datasets. Locus-­specific FST is plotted
against observed heterozygosity (Heterozygosity) with red circles indicating (a) 106 outlier SNPs under the hierarchical island model: SCS=
JR and the OT-­SB region= DK, HK, IR, IN, and OH, and (b) 138 outlier SNPs identified using Arlequin under the hierarchical island model:
S-­OT = DK and HK, M-­OT = IR and IN, and SB = OH with the three JR-­like individuals in the OH site of SB excluded. Yellow filled circles
represent the seven outliers found in both assumed models. Results of STRUCTURE analyse (c, d) and PCA (e, f) based on the outlier SNPs
identified in (a) and (b), respectively. Only one SNP per locus in each dataset, that is, 99 outlier SNPs identified in (a) and 125 outlier SNPs
identified in (b), was retained for STRUCTURE analyses and PCA to avoid bias derived from potential linkage disequilibrium

in introns, and seven (5.1%) in 3′-­UTRs. The outlier-­associated pro- 10.5%), transcription and translation (4, 10.5%), as well as those with
teins were manually classified into ten broad categories, including unknown functions (5, 13.2%) (Supporting Information Table S12).
carbohydrate (derivative) and lipid metabolism (3, 7.9%), cell death Although the number of candidate outliers was comparable under the
(1, 2.6%), DNA metabolism (3, 7.9%), localization (6, 15.8%), protein two assumed hierarchical island models, only seven SNPs were shared
folding, assembly, and metabolism (3, 7.9%), response to stimulus (3, between these two outlier SNP datasets (Figure 4a,b), indicating the two
7.9%), signaling (6, 15.8%), system development and processing (4, hierarchical island models relied on different genetic architecture.
 |
1924       XU et al.

3.6 | Genetic differentiation calculated based on the 3.8 | Signature of introgression in HK based on the

two outlier SNP datasets outlier SNP dataset
Values of pairwise FST calculated based on the first outlier SNP data- By using the USEPOPINFO model in STRUCTURE analyses based
set containing 106 outlier SNPs ranged from −0.0099 to 0.2785, on the second outlier SNP dataset containing 125 outlier SNPs (only
with statistical significance (p < 0.00001) detected in all pairwise one SNP per locus retained), eight individuals (53.3%) in the local
evaluations between the local population of JR and those in the OT-­ population of HK were identified to have mixed ancestry as either
SB region (FST range: 0.1953 to 0.2785) after Bonferroni correction migrants or hybrid descendants (Table 5). The consistency of results
(Supporting Information Table S13). based on different v values indicated the estimation to be robust.
Values of pairwise FST calculated based on the second outlier Values of h for each individual in the local population of HK
SNP dataset containing 138 outlier SNPs ranged from 0.0417 to computed by INTROGRESS based on the second outlier SNP data-
0.1964, with statistical significance (p < 0.00001) detected in all set containing 125 outlier SNPs (only one SNP per locus retained)
pairwise estimations between local populations in the OT-­SB re- ranged from 0.0373 to 0.7033 (Table 6). Among them, the eight in-
gion after Bonferroni correction (Supporting Information Table dividuals showing mixed ancestry revealed by STRUCTURE analyses
S14). Among them, values of pairwise FST calculated between the were estimated to have a moderate to large values of h ranging from
two local populations in the S-­OT (FST = 0.0685) as well as those 0.3643 to 0.7033, which indicated that these individuals were more
calculated between pairs of the three local populations in the M-­OT likely hybrid descendants rather than migrants.
and SB (FST range: 0.0417 to 0.0778) were smaller (Supporting
Information Table S14). In contrast, values of pairwise FST calcu-
3.9 | Directional relative migration patterns inferred
lated between the local populations in the S-­OT and those in the
based on the entire SNP dataset
M-­OT or SB (FST range: 0.0799 to 0.1964) were larger (Supporting
Information Table S14). Migration dynamic analyses based on the entire set of 5,458 SNPs
(only one SNP per locus retained) revealed extensive gene flow be-
tween local populations of B. platifrons, especially for those in the OT-­
3.7 | Population structure and individual assignment
SB region (Figure 5, Supporting Information Table S15). Although a
based on the two outlier SNP datasets
higher level of gene flow was detected between the local populations
STRUCTURE analyses based on the first outlier SNP dataset of JR and OH, gene flow between the local population of JR and the
containing 99 outlier SNPs (only one SNP per locus retained) others in the OT was found to be limited. Furthermore, gene flow
revealed two genetic groups (i.e., optimal K = 2) of B. platifrons from the local population of DK in the S-­OT to the others in the SCS,
in the Northwest Pacific (Supporting Information Figure S2c,d; the M-­OT, and SB was stronger than that in the reverse directions.
Figure 4c), which was in agreement with not only the result of
PCA based on the first outlier SNP dataset containing 99 outlier
SNPs (only one SNP per locus retained) along the first eigenvec- 4 | D I S CU S S I O N
tor (Figure 4e), but also the result of STRUCTURE analyses based
on the entire set of 5,458 SNPs (only one SNP per locus retained) Mitochondrial genes and genomewide SNPs were both applied in
(Figure 3a). the present study with the aim to have a deeper understanding of
STRUCTURE analyses based on the second outlier SNP data- the population genetics of the deep-­sea mussel B. platifrons in the
set containing 125 outlier SNPs (only one SNP per locus retained) Northwest Pacific. By using the three concatenated mitochondrial
uncovered two genetic groups (i.e., optimal K = 2) of B. platifrons genes, no significant genetic differentiation was detected between
in the OT-­S B region (Supporting Information Figure S2e,f), with any pairs of the six local populations. Meanwhile, the haplotype net-
one chiefly composed of individuals in the S-­OT and the other work constructed based on each mitochondrial gene also revealed
mainly composed of those in the M-­OT and SB (Figure 4d). no obvious genetic structure. These results were consistent with
However, two genetic backgrounds were detected in the local previous studies based on mitochondrial markers, indicating a high
population of HK in the S-­OT (Figure 4d). When using these out- dispersal capability and a lack of population differentiation of B. plat-
lier SNPs for PCA, all individuals in the S-­OT were clustered to- ifrons in the Northwest Pacific (Kyuno et al., 2009; Miyazaki et al.,
gether and formed a separate genetic group from those in the 2013; Shen et al., 2016).
M-­OT and SB along the first eigenvector (Figure 4f). Additionally, Analyses using the entire SNP dataset also indicated a high level
individuals in IR, IN, and OH appeared to form three small ge- of gene flow in B. platifrons manifesting by the small pairwise FST
netic groups along the second eigenvector (Figure 4f), which values and the result of migration dynamic analyses. Nevertheless,
was also observed in STRUCTURE analyses when forcing K = 4 by using the entire SNP dataset and the first outlier SNP dataset,
(Supporting Information Figure S3). significant genetic differentiation was uncovered between the local
XU et al. |
      1925

TA B L E   5   Ancestry inference for B.

q M-­OT and SB
platifrons in the local population of HK
inferred by STRUCTURE based on the Individual Prior pop v q prior pop Present Parent Grandparent
second outlier SNP dataset containing
HK_1 S-­OT 0.01 1 0 0 0
125 outlier SNPs (only one SNP per locus
retained). We estimated the posterior 0.05 1 0 0 0
probabilities (q) under three MIGRPRIOR 0.1 1 0 0 0
values (v) for each individual to have
HK_10 S-­OT 0.01 1 0 0 0
ancestry in the population group formed
by individuals in the S-­OT (i.e., q prior 0.05 1 0 0 0
pop), or in the population group formed by 0.1 1 0 0 0
those in the M-­OT and SB in the present HK_11 S-­OT 0.01 0 0.988 0.011 0
generation (present), in the first past
0.05 0 0.994 0.006 0
generation (parent), or the second past
generation (grandparent). Individuals in 0.1 0 0.996 0.004 0
bold indicate those which can be HK_12 S-­OT 0.01 0 0.737 0.26 0.003
considered as migrants or hybrid
0.05 0 0.834 0.165 0.002
descendants
0.1 0 0.883 0.116 0.001
HK_13 S-­OT 0.01 0 0.073 0.441 0.485
0.05 0 0.113 0.458 0.429
0.1 0 0.176 0.444 0.381
HK_14 S-­OT 0.01 0 0.061 0.811 0.128
0.05 0 0.082 0.814 0.104
0.1 0 0.104 0.807 0.089
HK_15 S-­OT 0.01 0.046 0 0.295 0.659
0.05 0.005 0 0.348 0.647
0.1 0.001 0 0.384 0.615
HK_2 S-­OT 0.01 0.997 0 0 0.002
0.05 0.987 0 0.001 0.011
0.1 0.973 0 0.003 0.024
HK_3 S-­OT 0.01 1 0 0 0
0.05 0.999 0 0 0.001
0.1 0.997 0 0 0.003
HK_4 S-­OT 0.01 0 0.094 0.708 0.198
0.05 0 0.118 0.709 0.173
0.1 0 0.132 0.708 0.16
HK_5 S-­OT 0.01 0.977 0 0 0.023
0.05 0.852 0 0 0.147
0.1 0.684 0 0.001 0.315
HK_6 S-­OT 0.01 1 0 0 0
0.05 0.999 0 0 0.001
0.1 0.998 0 0 0.002
HK_7 S-­OT 0.01 1 0 0 0
0.05 0.999 0 0 0.001
0.1 0.999 0 0 0.001
HK_8 S-­OT 0.01 0 0.486 0.401 0.113
0.05 0 0.57 0.355 0.076
0.1 0 0.636 0.306 0.058
HK_9 S-­OT 0.01 0.188 0 0.075 0.738
0.05 0.022 0 0.098 0.879
0.1 0.006 0 0.107 0.887
 |
1926       XU et al.

TA B L E   6   The hybrid index (h) computed by INTROGRESS for B.

platifrons in the local population of HK based on the second outlier
SNP dataset containing 125 outlier SNPs (only one SNP per locus
0.64
retained)
0.79
Individual h 95% CI 0.69
HK_1 0.0373 (0.0020, 0.1624)
0.57 0.64
HK_10 0.1006 (0.0173, 0.2797)
1.00
HK_11 0.6874 (0.4703, 0.9172)
HK_12 0.7033 (0.4607, 0.9454)
0.71 0.81
HK_13 0.5302 (0.3161, 0.783)
0.78
HK_14 0.6013 (0.3712, 0.8596) 0.52
HK_15 0.4886 (0.2734, 0.7555)
0.64 0.87
0.52
HK_2 0.3591 (0.0642, 0.8843) 0.56 0.52
HK_3 0.1629 (0.0407, 0.3972) 0.68
HK_4 0.5987 (0.3719, 0.8548)
0.50 0.71
HK_5 0.2217 (0.0746, 0.4476)
HK_6 0.0995 (0.0171, 0.2776)
0.78 0.58
HK_7 0.0916 (0.0150, 0.2699)
HK_8 0.6574 (0.4292, 0.9071)
HK_9 0.3643 (0.1588, 0.6334)

Notes. CI: confidence interval.

Individuals in bold indicate those which can be considered as migrants or
hybrid descendants by using STRUCTURE. F I G U R E   5   Directional relative migration of B. platifrons estimated
using divMigrate-­online based on the entire set of 5,458 SNPs (only
one SNP per locus retained) between sampling locations. Arrows
population in the SCS and those in the OT-­SB region, and a clear ge- refer to the direction of gene flow. Different colors of the circles
indicate the geological regions of the sampling locations. Arrows
netic subdivision was detected between individuals in the SCS and
with larger numbers display thicker in shape and darker in color.
those in the OT-­SB region. Moreover, by using the entire SNP data-
Numbers on the arrows represent the relative migration coefficients
set and the second outlier SNP dataset, a minor genetic divergence derived from GST statistics. Only the coefficients larger than 0.50 are
was uncovered between individuals in the S-­OT and those in the indicated, and the complete relative directional migration matrix is
M-­OT and SB. shown in Supporting Information Table S15
The discrepancy between results obtained from analyzing mito-
chondrial genes and SNPs was considered to be due to the different
number and types of genetic markers used, resulting in different differentiation between local populations of B. platifrons as revealed
resolution. It has been reported that analyses of only a single or a by the mitochondrial genes. Therefore, we focused our discussion
few genetic markers can lead to relatively larger confidence inter- below on the results obtained by using genomewide SNPs.
vals for very small values of FST, which may result in non-­statistical
significance for species with extensive gene flow (Blanco-­Bercial
4.1 | Co-­occurrence of two cryptic semi-­isolated
& Bucklin, 2016). Furthermore, it has been shown that selective
lineages of B. platifrons in the Northwest Pacific
sweeps at mitochondrial loci might be more common for organisms
living in chemosynthesis-­based ecosystems in the deep ocean when Oceanographic connection between the semi-­enclosed marginal
compared with those in other marine environments (Roterman, SCS and the Northwest Pacific is mainly achieved through the
Copley, Linse, Tyler, & Rogers, 2016). The significantly negative seasonal intrusions of the Kuroshio Current, the North Pacific
values of Tajima’s D and Fu’s FS statistics estimated based on the Intermediate Water (NPIW), as well as the Pacific Deep Water
mitochondrial genes of B. platifrons indicated a signature of selec- through the Luzon Strait (Nan et al., 2011; Qu, Girton, & Whitehead,
tive sweeps at these mitochondrial genes or population expansion 2006; You et al., 2005). The Kuroshio Current is a dominant and
after a recent bottleneck (Fu, 1997; Tajima, 1989). However, the dual strong warm surface current in the Northwest Pacific that origi-
analyses of Tajima’s D and Fu and Li’s D* statistics on SNP markers nates from the North Equatorial Current and runs northeastward
paved an argument for selective sweeps at the mitochondrial genes, along the Philippines coast (Andres et al., 2015). When passing by
as only a few SNP markers agreed with the scenario of population the Luzon Strait, a branch of the Kuroshio water can intrude into
expansion. In addition, the scenario of selective sweeps at the mi- the SCS in a strong anticyclonic circulation, which is known as the
tochondrial genes also fitted well with the observed lack of genetic “looping path,” and then flows out of the SCS to join its mainstream
XU et al. |
      1927

that continues to run toward the eastern Taiwan, the outer shelf (Ikegami, Tsuji, Kumagai, Ishibashi, & Takai, 2015). These geological
of the East China Sea, passing by SB, and extending to the south- settings indicate the existence of subtle geographic barriers be-
eastern Japan (Figure 1b; Andres et al., 2015; Nan et al., 2011). The tween the S-­OT and the M-­OT, which may decrease the dispersal
NPIW is a mid-­depth water mass widely distributed in the North success of B. platifrons larvae across different regions of the OT.
Pacific subtropical gyre, which can flow from the northeastern In addition, the ancestry inference of STRUCTURE analyses and
part of North Pacific into the SCS, especially during winter and the moderate to large values of h revealed that eight individuals in
spring (You et al., 2005). These ocean currents make it possible for the local population of HK to be hybrid descendants of individuals
individuals of B. platifrons on the two sides of the Luzon Strait to in the S-­OT mating with those originating from the M-­OT and SB.
exchange larvae. However, as revealed by the result of migration This result indicated that HK may represent a contact zone for larval
dynamic analyses, such exchange tended to be limited due to the dispersal and genetic exchange of B. platifrons in the OT-­SB region.
relatively small volume of seawater involved. Therefore, the Luzon Individuals of B. platifrons in HK inhabited the deepest vent field in-
Strait may have served as a dispersal barrier, either promoting the cluded in this study, and this vent field is contained within a caldera.
formation of two cryptic semi-­isolated lineages of B. platifrons (i.e., Therefore, it is possible that the introgression signature observed
the JR lineage and the OT-­SB lineage) in the Northwest Pacific or here was related to such bathymetry and topography, which may
trapping a preexisting genetic barrier. Similar situations have also serve to trap mussel larvae from different genetic groups.
been reported from other Bathymodiolus mussels, such as those
from the mid-­Atlantic Ridge (e.g., Breusing, Vrijenhoek, & Reusch,
4.3 | Directional migration patterns of B. platifrons
2017; Breusing et al., 2016; Faure, Jollivet, Tanguy, Bonhomme, &
in the Northwest Pacific
Bierne, 2009; O’Mullan, Maas, Lutz, & Vrijenhoek, 2001) and the
East Pacific (e.g., Plouviez et al., 2013). The result of migration dynamic analyses revealed that gene flow
Intriguingly, the results of STRUCTURE analyses and PCA re- of B. platifrons in the Northwest Pacific to be asymmetrical, in that
vealed that three individuals in the OH site of SB had an extremely the gene flow was stronger from the S-­OT (especially DK) toward
high genetic similarity to those dominating the JR site in the SCS, de- the SCS, the M-­OT, and SB than that in the reverse directions. Such
spite no such individuals being found in any sampled vent fields in the migration patterns indicated that the local populations in the S-­OT
OT that lies between SB and the SCS. This phenomenon might be a tended to be the source populations of B. platifrons in the Northwest
consequence of long-­distance migration events between JR and OH Pacific, which was in agreement with the lower values of Hobs and
as revealed by the result of migration dynamic analyses. Another hy- π as well as higher values of FIS exhibited by the local populations
pothesis is that the two cryptic semi-­isolated lineages of B. platifrons in the S-­OT compared to those of the other local populations.
may have a difference in habitat preference as it has been observed in However, this deduction should be treated with caution, since addi-
coastal mussels (Bierne, Bonhomme, & David, 2003), with the lineage tional populations of B. platifrons elsewhere in the Northwest Pacific
dominating JR preferring methane seeps. Nevertheless, broader sam- likely remain unsampled.
pling is required in the future to better test this hypothesis.

5 | CO N C LU S I O N S A N D PE R S PEC TI V E S
4.2 | Genetic homogeneity, fine-­scale genetic
structure, and admixture of B. platifrons in the OT-­
By using genomewide SNPs rather than mitochondrial genes, two
SB region
cryptic semi-­isolated lineages of B. platifrons in the Northwest Pacific
The absence of genetic differentiation for B. platifrons in the OT-­SB were identified in this study, which may have been formed due to the
region was revealed by the nonsignificant pairwise FST values and barrier effect of the Luzon Strait or the contact zone been trapped
the result of STRUCTURE analyses based on the entire SNP data- by it. Among them, one lineage is mainly distributed in the semi-­
set, indicating that the mainstream of the Kuroshio Current and the enclosed marginal SCS, while the other is mainly distributed across
NPIW have played a vital role in promoting their larval dispersal in the OT-­SB region. In addition, a fine-­scale population structure was
this area. Nevertheless, carrying out the above analyses and PCA detected for B. platifrons in the OT-­SB region, which might be due to
using the second outlier SNP dataset, individuals in the S-­OT were the existence of subtle geographic barriers between the S-­OT and
detected to be genetically separated from those in the M-­OT and SB. the M-­OT. The occurrence of three individuals with a high genetic
Such fine-­scale population genetic structure could be explained as a affinity to those in JR in the OH site of SB remains puzzling and war-
result of natural selection in response to local adaptation (Gagnaire rants further investigation to test whether this is related to differ-
et al., 2015; Milano et al., 2014). Furthermore, the topography of the ences in habitat preferences (i.e., methane seeps vs hydrothermal
OT may have also played a key role in shaping the observed popula- vents). The mixed genetic backgrounds detected in the local popu-
tion subdivision. The OT is a back-­arc rifting basin formed behind lation of HK in the S-­OT indicated that this area may represent a
the Ryukyu trench-­arc system. The M-­OT is located at the transi- contact zone for larvae from different locations in the OT-­SB region.
tional region between the deeper S-­OT and the shallower northern Moreover, the local populations in the S-­OT (especially DK) may
OT, which is associated with the occurrence of intra-­trough grabens serve as a potential source of B. platifrons in the Northwest Pacific.
 |
1928       XU et al.

Overall, the present study has enhanced our understanding of ORCID

the genetic connectivity, fine-­scale genetic structure, and migra-
Chong Chen  http://orcid.org/0000-0002-5035-4021
tion patterns of B. platifrons. Together with several recent studies
(Breusing et al., 2016, 2017), this study exemplified the usefulness Jian-Wen Qiu  http://orcid.org/0000-0002-1541-9627

of SNP data especially from high-­throughput sequencing for pop-

ulation genetic studies of chemosynthetic ecosystems. More im-
REFERENCES
portantly, the results from this study will lay a foundation for an
effective determination of biogeographic regions, establishment of Andres, M., Jan, S., Sanford, T. B., Mensah, V., Centurioni, L. R., & Book,
J. W. (2015). Mean structure and variability of the Kuroshio from
informed management plans, and designation of deep-­sea reserves
northeastern Taiwan to southwestern Japan. Oceanography, 28, 84–
in the Pacific Ocean, in preparation for an upcoming era of deep-­sea 95. https://doi.org/10.5670/oceanog
resource exploitation (Miller et al., 2018; Sigwart, Chen, & Marsh, Arellano, S. M., Van Gaest, A. L., Johnson, S. B., Vrijenhoek, R. C., &
2017). Young, C. M. (2014). Larvae from deep-­sea methane seeps disperse in
surface waters. Proceedings of the Royal Society of London B: Biological
Sciences, 281, 20133276. https://doi.org/10.1098/rspb.2013.3276
AC K N OW L E D G E M E N T S Arèvalo, E., Davis, S. K., & Sites, J. W. Jr (1994). Mitochondrial DNA
sequence divergence and phylogenetic relationships among
We thank the captains and crews of the R/Vs Natsushima, Kaiyo, and eight chromosome races of the Sceloporus grammicus complex
(Phrynosomatidae) in central Mexico. Systematic Biology, 43, 387–
Xiangyanghong 9, as well as the pilots and operation teams of the
418. https://doi.org/10.1093/sysbio/43.3.387
manned deep-­submergence vehicle Jiaolong and the ROV Hyper- Baco, A. R., Etter, R. J., Ribeiro, P. A., Heyden, S., Beerli, P., & Kinlan, B.
Dolphin for sampling during the relevant cruises. We thank the P. (2016). A synthesis of genetic connectivity in deep-­sea fauna and
principal scientists of the relevant research cruises: Dr. Katsunori implications for marine reserve design. Molecular Ecology, 25, 3276–
3298. https://doi.org/10.1111/mec.13689
Fujikura (JAMSTEC) for the R/V Natsushima cruise NT09-­0 6 leg
Bielawski, J. P., & Gold, J. R. (1996). Unequal synonymous substitu-
1, Dr. Shinsuke Kawagucci (JAMSTEC) for the R/V Kaiyo cruise tion rates within and between two protein-­coding mitochondrial
KY14-­02, Dr. Hiroshi Miyake (Kitasato University) for the R/V genes. Molecular Biology and Evolution, 13, 889–892. https://doi.
Kaiyo cruise KY11-­02 leg 2, Dr. Hidetaka Nomaki (JAMSTEC) for org/10.1093/oxfordjournals.molbev.a025649
the R/V Kaiyo cruise KY11-­01 leg1, Dr. Ken Takai (JAMSTEC) for Bierne, N., Bonhomme, F., & David, P. (2003). Habitat preference and the
marine-­speciation paradox. Proceedings of the Royal Society of London
the R/V Kaiyo cruise KY14-­01, Dr. Hiroyuki Yamamoto (JAMSTEC)
B: Biological Sciences, 270, 1399–1406. https://doi.org/10.1098/
for the R/V Natsushima cruise NT13-­22, as well as Dr. Feng Liu rspb.2003.2404
(COMRA) and Dr. Huaiyang Zhou (Tongji University) for the R/V Blanco-Bercial, L., & Bucklin, A. (2016). New view of population ge-
Xiangyanghong 9 cruise Dayang-­31. We also thank Dr. Tadashi netics of zooplankton: RAD-­s eq analysis reveals population struc-
ture of the North Atlantic planktonic copepod Centropages typi-
Maruyama (JAMSTEC) for his help in establishing collaborations
cus. Molecular Ecology, 25, 1566–1580. https://doi.org/10.1111/
among the co-­authors. mec.13581
Breusing, C., Biastoch, A., Drews, A., Metaxas, A., Jollivet, D., Vrijenhoek,
R. C., … Reusch, T. B. (2016). Biophysical and population genetic mod-
DATA A R C H I V I N G S TAT E M E N T els predict the presence of “phantom” stepping stones connecting
Mid-­Atlantic Ridge vent ecosystems. Current Biology, 26, 2257–2267.
The mitochondrial gene sequences were deposited in GenBank of https://doi.org/10.1016/j.cub.2016.06.062
National Center for Biotechnology Information (NCBI) under the ac- Breusing, C., Vrijenhoek, R. C., & Reusch, T. B. (2017). Widespread intro-
cession numbers of MH389991 to MH390100 for cox1, MH390101 gression in deep-­sea hydrothermal vent mussels. BMC Evolutionary
Biology, 17, 13. https://doi.org/10.1186/s12862-016-0862-2
to MH390210 for nad4, and MH390211 to MH390320 for atp6.
Bruno, J. F., & Bertness, M. D. (2001). Habitat modification and facil-
Raw reads of 2b-­R AD libraries were deposited in the Sequence itation in benthic marine communities. In M. D. Bertness, S. D.
Read Archive (SRA) database of NCBI under the accession number Gaines, & M. E. Hay (Eds.), Marine community ecology (pp. 201–218).
of SRP149310. Sunderland, MA: Sinauer.
Buerkle, C. A. (2005). Maximum-­likelihood estimation of a hybrid index
based on molecular markers. Molecular Ecology Notes, 5, 684–687.
AU T H O R S ’ C O N T R I B U T I O N S https://doi.org/10.1111/j.1471-8286.2005.01011.x
Catchen, J., Hohenlohe, P. A., Bassham, S., Amores, A., & Cresko, W. A.
JWQ, PYQ, TX, and JS conceived this project. JWQ, JS, HKW, CC, (2013). Stacks: An analysis tool set for population genomics. Molecular
MN, and DF collected the samples. TX extracted DNA, conducted Ecology, 22, 3124–3140. https://doi.org/10.1111/mec.12354
Cowen, R. K., & Sponaugle, S. (2009). Larval dispersal and marine pop-
2b-­R AD libraries under the guidance of JL, SW, and ZB, performed
ulation connectivity. Annual Review of Marine Science, 1, 443–466.
bioinformatics analyses, and drafted the manuscript. RJ helped in- https://doi.org/10.1146/annurev.marine.010908.163757
terpret the roles of ocean currents in population connectivity among Earl, D. A., & vonHoldt, B. M. (2012). STRUCTURE HARVESTER: A web-
mussel populations. All authors contributed to improvement of the site and program for visualizing STRUCTURE output and implement-
ing the Evanno method. Conservation Genetics Resources, 4, 359–361.
manuscript.
https://doi.org/10.1007/s12686-011-9548-7
XU et al. |
      1929

Edgar, R. C. (2004). MUSCLE: Multiple sequence alignment with high ac- Kimura, M. (1980). A simple method for estimating evolutionary rates
curacy and high throughput. Nucleic Acids Research, 32, 1792–1797. of base substitutions through comparative studies of nucleotide
https://doi.org/10.1093/nar/gkh340 sequences. Journal of Molecular Evolution, 16, 111–120. https://doi.
Excoffier, L., Hofer, T., & Foll, M. (2009). Detecting loci under selection org/10.1007/BF01731581
in a hierarchically structured population. Heredity, 103, 285–298. Kinlan, B. P., & Gaines, S. D. (2003). Propagule dispersal in marine and
https://doi.org/10.1038/hdy.2009.74 terrestrial environments: A community perspective. Ecology, 84,
Excoffier, L., & Lischer, H. E. (2010). Arlequin suite ver 3.5: A new series 2007–2020. https://doi.org/10.1890/01-0622
of programs to perform population genetics analyses under Linux Kumar, S., Stecher, G., & Tamura, K. (2016). MEGA7: Molecular
and Windows. Molecular Ecology Resources, 10, 564–567. https://doi. Evolutionary Genetics Analysis version 7.0 for bigger data-
org/10.1111/j.1755-0998.2010.02847.x sets. Molecular Biology and Evolution, 33, 1870–1874. https://doi.
Falush, D., Stephens, M., & Pritchard, J. K. (2007). Inference of popu- org/10.1093/molbev/msw054
lation structure using multilocus genotype data: Dominant markers Kyuno, A., Shintaku, M., Fujita, Y., Matsumoto, H., Utsumi, M., Watanabe,
and null alleles. Molecular Ecology Notes, 7, 574–578. https://doi. H., … Miyazaki, J. I. (2009). Dispersal and differentiation of deep-­sea
org/10.1111/j.1471-8286.2007.01758.x mussels of the genus Bathymodiolus (Mytilidae, Bathymodiolinae).
Faure, B., Jollivet, D., Tanguy, A., Bonhomme, F., & Bierne, N. (2009). Journal of Marine Biology, 2009, 625672.
Speciation in the deep sea: Multi-­locus analysis of divergence and Le Bris, N., Arnaud-Haond, S., Beaulieu, S., Cordes, E., Hilario, A., Rogers,
gene flow between two hybridizing species of hydrothermal vent A., … Watanabe, H. (2016). Hydrothermal vents and cold seeps.
mussels. PLoS ONE, 4, e6485. https://doi.org/10.1371/journal. In The First Global Integrated Marine Assessment: World Ocean
pone.0006485 Assessment I. United Nations, New York.
Folmer, O., Black, M., Hoeh, W., Lutz, R., & Vrijenhoek, R. (1994). DNA Leigh, J. W., & Bryant, D. (2015). POPART: Full-­feature software for hap-
primers for amplification of mitochondrial cytochrome C oxidase lotype network construction. Methods in Ecology and Evolution, 6,
subunit I from diverse metazoan invertebrates. Molecular Marine 1110–1116. https://doi.org/10.1111/2041-210X.12410
Biology and Biotechnology, 3, 294–299. Levin, L. A., Baco, A. R., Bowden, D. A., Colaco, A., Cordes, E. E.,
Fu, Y. X. (1997). Statistical tests of neutrality of mutations against pop- Cunha, M. R., … Watling, L. (2016). Hydrothermal vents and meth-
ulation growth, hitchhiking and background selection. Genetics, 147, ane seeps: Rethinking the sphere of influence. Frontiers in Marine
915–925. Science, 3, 72.
Fu, Y. X., & Li, W. H. (1993). Statistical tests of neutrality of mutations. Li, R., Yu, C., Li, Y., Lam, T. W., Yiu, S. M., Kristiansen, K., & Wang, J.
Genetics, 133, 693–709. (2009). SOAP2: An improved ultrafast tool for short read align-
Fujikura, K., Tsuchida, S., Nunoura, T., Soh, W., Machiyama, H., Huang, ment. Bioinformatics, 25, 1966–1967. https://doi.org/10.1093/
C., & Lin, S. (2007). Vent-type chemosynthetic community associ- bioinformatics/btp336
ated with methane seep at the Formosa Ridge, off southwest Taiwan. Lischer, H. E., & Excoffier, L. (2012). PGDSpider: An automated data
Proceedings of the International Conference on Gas Hydrate: Energy, conversion tool for connecting population genetics and genomics
Climate and Environment, Taipei. programs. Bioinformatics, 28, 298–299. https://doi.org/10.1093/
Gagnaire, P. A., Broquet, T., Aurelle, D., Viard, F., Souissi, A., Bonhomme, F., bioinformatics/btr642
… Bierne, N. (2015). Using neutral, selected, and hitchhiker loci to assess Lorion, J., Kiel, S., Faure, B., Kawato, M., Ho, S. Y., Marshall, B., … Fujiwara,
connectivity of marine populations in the genomic era. Evolutionary Y. (2014). Adaptive radiation of chemosymbiotic deep-­sea mussels.
Applications, 8, 769–786. https://doi.org/10.1111/eva.12288 Proceedings of the Royal Society B: Biological Sciences, 281, 20132972.
Gompert, Z., & Alex Buerkle, C. (2010). INTROGRESS: A soft- Manoukis, N. C. (2007). FORMATOMATIC: A program for convert-
ware package for mapping components of isolation in hy- ing diploid allelic data between common formats for population
brids. Molecular Ecology Resources, 10, 378–384. https://doi. genetic analysis. Molecular Ecology Notes, 7, 592–593. https://doi.
org/10.1111/j.1755-0998.2009.02733.x org/10.1111/j.1471-8286.2007.01784.x
Govenar, B. (2010). Shaping vent and seep communities: Habitat provi- McVeigh, D. M., Eggleston, D. B., Todd, A. C., Young, C. M., & He, R.
sion and modification by foundation species. In S. Kiel (Ed.), The vent (2017). The influence of larval migration and dispersal depth on po-
and seep biota: Aspects from microbes to ecosystems (pp. 403–432). tential larval trajectories of a deep-­sea bivalve. Deep Sea Research
Dordrecht, the Netherlands: Springer. Part I: Oceanographic Research Papers, 127, 57–64. https://doi.
Hohenlohe, P. A., Amish, S. J., Catchen, J. M., Allendorf, F. W., & Luikart, org/10.1016/j.dsr.2017.08.002
G. (2011). Next-­generation RAD sequencing identifies thousands of Milano, I., Babbucci, M., Cariani, A., Atanassova, M., Bekkevold, D.,
SNPs for assessing hybridization between rainbow and westslope Carvalho, G. R., … Bargelloni, L. (2014). Outlier SNP markers reveal
cutthroat trout. Molecular Ecology Resources, 11, 117–122. https:// fine-­scale genetic structuring across European hake populations
doi.org/10.1111/j.1755-0998.2010.02967.x (Merluccius merluccius). Molecular Ecology, 23, 118–135. https://doi.
Ikegami, F., Tsuji, T., Kumagai, H., Ishibashi, J. I., & Takai, K. (2015). org/10.1111/mec.12568
Active rifting structures in Iheya Graben and adjacent area of the Miller, K. A., Thompson, K. F., Johnston, P., & Santillo, D. (2018). An over-
Mid-Okinawa Trough observed through seismic reflection surveys. view of seabed mining including the current state of development,
In J. Ishibashi, K. Okino, & M. Sunamura (Eds.), Subseafloor bio- environmental impacts and knowledge gaps. Frontiers in Marine
sphere linked to hydrothermal systems (pp. 361–368). Tokyo, Japan: Science, 4, 418. https://doi.org/10.3389/fmars.2017.00418
Springer. Miyazaki, J. I., Beppu, S., Kajio, S., Dobashi, A., Kawato, M., Fujiwara, Y., &
Jakobsson, M., & Rosenberg, N. A. (2007). CLUMPP: A cluster match- Hirayama, H. (2013). Dispersal ability and environmental adaptabil-
ing and permutation program for dealing with label switching and ity of deep-­sea mussels Bathymodiolus (Mytilidae: Bathymodiolinae).
multimodality in analysis of population structure. Bioinformatics, 23, Open Journal of Marine Science, 3, 31–39. https://doi.org/10.4236/
1801–1806. https://doi.org/10.1093/bioinformatics/btm233 ojms.2013.31003
Jiao, W., Fu, X., Dou, J., Li, H., Su, H., Mao, J., … Bao, Z. (2014). High-­ MolluscaBase (2018). Bathymodiolus Kenk & B. R. Wilson, 1985. Accessed
resolution linkage and quantitative trait locus mapping aided by ge- through: World Register of Marine Species at: http://marinespecies.
nome survey sequencing: Building up an integrative genomic frame- org/aphia.php?p=taxdetails&id=138214 on 2018-04-26.
work for a bivalve mollusc. DNA Research, 21, 85–101. https://doi. Nan, F., Xue, H., Chai, F., Shi, L., Shi, M., & Guo, P. (2011). Identification
org/10.1093/dnares/dst043 of different types of Kuroshio intrusion into the South China
 |
1930       XU et al.

Sea. Ocean Dynamics, 61, 1291–1304. https://doi.org/10.1007/ Ecology and Evolution, 6, 3461–3475. https://doi.org/10.1002/
s10236-011-0426-3 ece3.2096
Nei, M. (1973). Analysis of gene diversity in subdivided populations. Tajima, F. (1989). Statistical method for testing the neutral mutation hy-
Proceedings of the National Academy of Sciences of the United States of pothesis by DNA polymorphism. Genetics, 123, 585–595.
America, 70, 3321–3323. https://doi.org/10.1073/pnas.70.12.3321 Tunnicliffe, V., Juniper, S. K., & Sibuet, M. (2003). Reducing environments
O’Mullan, G. D., Maas, P. A., Lutz, R. A., & Vrijenhoek, R. C. (2001). A of the deep-sea floor. In P. A. Tyler (Ed.), Ecosystems of the deep oceans
hybrid zone between hydrothermal vent mussels (Bivalvia: Mytilidae) (pp. 81–110). Amsterdam, the Netherlands: Elsevier.
from the Mid-­Atlantic Ridge. Molecular Ecology, 10, 2819–2831. Vaidya, G., Lohman, D. J., & Meier, R. (2011). SequenceMatrix:
https://doi.org/10.1046/j.0962-1083.2001.01401.x Concatenation software for the fast assembly of multi-­gene datasets
Plouviez, S., Faure, B., Le Guen, D., Lallier, F. H., Bierne, N., & Jollivet, with character set and codon information. Cladistics, 27, 171–180.
D. (2013). A new barrier to dispersal trapped old genetic clines that https://doi.org/10.1111/j.1096-0031.2010.00329.x
escaped the Easter Microplate tension zone of the Pacific vent Van Dover, C. L. (2000). The ecology of deep-sea hydrothermal vents.
mussels. PLoS ONE, 8, e81555. https://doi.org/10.1371/journal. Princeton, NJ: Princeton University Press.
pone.0081555 Vrijenhoek, R. C. (2010). Genetic diversity and connectivity of deep-­sea
Pritchard, J. K., Stephens, M., & Donnelly, P. (2000). Inference of pop- hydrothermal vent metapopulations. Molecular Ecology, 19, 4391–
ulation structure using multilocus genotype data. Genetics, 155, 4411. https://doi.org/10.1111/j.1365-294X.2010.04789.x
945–959. Wang, S., Meyer, E., McKay, J. K., & Matz, M. V. (2012). 2b-­R AD: A simple
Qu, T., Girton, J. B., & Whitehead, J. A. (2006). Deepwater overflow and flexible method for genome-­wide genotyping. Nature Methods, 9,
through Luzon strait. Journal of Geophysical Research, 111, C01002. 808–810. https://doi.org/10.1038/nmeth.2023
Roesti, M., Salzburger, W., & Berner, D. (2012). Uninformative Watanabe, H., Fujikura, K., Kojima, S., Miyazaki, J. I., & Fujiwara, Y. (2010).
polymorphisms bias genome scans for signatures of se- Japan: Vents and seeps in close proximity. In S. Kiel (Ed.), The vent
lection. BMC Evolutionary Biology, 12, 94. https://doi. and seep biota (pp. 379–401). Dordrecht, the Netherlands: Springer.
org/10.1186/1471-2148-12-94 https://doi.org/10.1007/978-90-481-9572-5
Rogers, A. D., Tyler, P. A., Connelly, D. P., Copley, J. T., James, R., Larter, Xu, T., Sun, J., Lv, J., Watanabe, H. K., Li, T., Zou, W., … Qiu, J. W. (2017).
R. D., … Zwirglmaier, K. (2012). The discovery of new deep-­sea Genome-­wide discovery of single nucleotide polymorphisms (SNPs)
hydrothermal vent communities in the Southern Ocean and impli- and single nucleotide variants (SNVs) in deep-­sea mussels: Potential
cations for biogeography. PLoS Biology, 10, e1001234. https://doi. use in population genomics and cross-­species application. Deep
org/10.1371/journal.pbio.1001234 Sea Research Part II: Topical Studies in Oceanography, 137, 318–326.
Rosenberg, N. A. (2004). DISTRUCT: A program for the graphical display https://doi.org/10.1016/j.dsr2.2016.03.011
of population structure. Molecular Ecology Notes, 4, 137–138. You, Y., Chern, C. S., Yang, Y., Liu, C. T., Liu, K. K., & Pai, S. C. (2005). The
Roterman, C. N., Copley, J. T., Linse, K. T., Tyler, P. A., & Rogers, A. South China Sea, a cul-de-sac of North Pacific intermediate water.
D. (2016). Connectivity in the cold: The comparative population Journal of Oceanography, 61, 509–527. https://doi.org/10.1007/
genetics of vent-­e ndemic fauna in the Scotia Sea, Southern Ocean. s10872-005-0059-6
Molecular Ecology, 25, 1073–1088. https://doi.org/10.1111/ Young, C. M., He, R., Emlet, R. B., Li, Y., Qian, H., Arellano, S. M., … Rice,
mec.13541 M. E. (2012). Dispersal of deep-­sea larvae from the intra-­American
Rozas, J., Sánchez-DelBarrio, J. C., Messeguer, X., & Rozas, R. (2003). seas: Simulations of trajectories using ocean models. Integrative and
DnaSP, DNA polymorphism analyses by the coalescent and other Comparative Biology, 52, 483–496. https://doi.org/10.1093/icb/
methods. Bioinformatics, 19, 2496–2497. https://doi.org/10.1093/ ics090
bioinformatics/btg359 Zheng, X., Levine, D., Shen, J., Gogarten, S. M., Laurie, C., & Weir, B. S.
Sambrook, J., Fritsch, E. F., & Maniatis, T. (1989). Molecular cloning: A lab- (2012). A high-­performance computing toolset for relatedness and
oratory manual (2nd ed.). Cold Spring Harbor, NY: Cold Spring Harbor principal component analysis of SNP data. Bioinformatics, 28, 3326–
Laboratory Press. 3328. https://doi.org/10.1093/bioinformatics/bts606
Shen, Y., Kou, Q., Chen, W., He, S., Yang, M., Li, X., & Gan, X. (2016).
Comparative population structure of two dominant species, Shinkaia
crosnieri (Munidopsidae: Shinkaia) and Bathymodiolus platifrons S U P P O R T I N G I N FO R M AT I O N
(Mytilidae: Bathymodiolus), inhabiting both deep-­sea vent and cold
seep inferred from mitochondrial multi-­genes. Ecology and Evolution, Additional supporting information may be found online in the
6, 3571–3582. https://doi.org/10.1002/ece3.2132 Supporting Information section at the end of the article.     
Sigwart, J. D., Chen, C., & Marsh, L. (2017). Is mining the seabed bad for
mollusks? The Nautilus, 131, 43–49.
Suess, E. (2005). RV Sonne cruise report SO 177: SiGer 2004; Sino-German
How to cite this article: Xu T, Sun J, Watanabe HK, et al.
Cooperative Project; South China Sea continental margin: geological
methane budget and environmental effects of methane emissions and Population genetic structure of the deep-­sea mussel
gashydrates. Bathymodiolus platifrons (Bivalvia: Mytilidae) in the Northwest
Sun, J., Zhang, Y., Xu, T., Zhang, Y., Mu, H., Zhang, Y., … Qian, P. Y. (2017). Pacific. Evol Appl. 2018;11:1915–1930. https://doi.
Adaptation to deep-­sea chemosynthetic environments as revealed
org/10.1111/eva.12696
by mussel genomes. Nature Ecology & Evolution, 1, 0121. https://doi.
org/10.1038/s41559-017-0121
Sundqvist, L., Keenan, K., Zackrisson, M., Prodöhl, P., & Kleinhans, D.
(2016). Directional genetic differentiation and relative migration.