Postglacial Lineage Admixture in the Contact Zones of the Two Japanese Deciduous Broad-leaved Tree Species Estimated by Nuclear Microsatellite and Chloroplast DNA Markers

2016·0.84 MB·English

by  Tono, Akitaka

The Japanese Society for Plant Systematics
ISSN 1346-7565 Acta Phytotax. Geobot. 67 (1): 1–16 (2016)
Postglacial Lineage Admixture in the Contact Zones of the Two Japanese
Deciduous Broad-leaved Tree Species Estimated by Nuclear Microsatellite
and Chloroplast DNA Markers
AkitAkA tono1,*, tAkAyA iwAsAki2, Akihiro seo3,† And noriAki MurAkAMi1
1 Makino Herbarium, Tokyo Metropolitan University, 1-1 Minamiosawa, Hachioji, Tokyo 192-0397, Japan.
*[email protected] (author for correspondence); 2Center for Ecological Research,
Kyoto University, 2-509-3, Hirano, Otsu, Shiga 520-2113, Japan; 3Department of Botany, Graduate School of Science,
Kyoto University, Kitashirakawa-Oiwake-cho, Sakyo-ku, Kyoto 606-8502, Japan; †Present address; Kochi Prefectur-
al Makino Botanical Garden, 4200-6, Godaisan, Kochi, Kochi 781-8125, Japan
When historically isolated populations meet during postglacial expansion, a mixed distribution of dis-
tinct DNA lineages called contact zones is created. The gradual dissolution of the spatial genetic struc-
tures in contact zones should be related to differences in pollen and seed dispersal, given no restriction
on gene flow by e.g. reproductive isolation. We aimed to clarify effects of pollen dispersal modes on
nuclear DNA (nrDNA) genetic structures of two codistributed species with different pollen dispersal
modes, by analyzing nuclear microsatellites of the insect-pollinated Magnolia obovata and the wind-
pollinated Carpinus laxiflora, which show highly consistent contact zone locations in terms of chloro-
plast DNA (cpDNA). The genetic structure based on the nrDNA and that based on the cpDNA were con-
cordant in M. obovata, but not so concordant in C. laxiflora. Pollen dispersal ability is higher in the wind-
pollinated C. laxiflora than in M. obovata, resulting in the higher estimated pollen/seed migration ratio
in C. laxiflora than in M. obovata. Therefore, the extent of postglacial lineage admixture in nrDNA was
predominant in C. laxiflora. Our results suggested that differences in pollen dispersal ability may affect
the nrDNA genetic structure between co-distributed species with common migration histories in the
same area.
Key words: admixture, contact zone, deciduous broad-leaved forest, gene flow, microsatellites, phylo-
geographic studies
Climate oscillations during the Quaternary 2002, McLachlan et al. 2005, Magri et al. 2006).
affected geographical distribution of various Contact zones are defined as areas where isolated
plant and animal species (Hewitt 2004). During populations from different refugia meet during
climatic oscillations, the ranges of organisms postglacial expansion and distinct DNA lineages
might have occasionally been divided into sever- are consequently able to mix (Harrison 1993,
al isolated populations. The genetic structure in Taberlet et al. 1998). Although secondary con-
extant populations may be the result of such his- tacts of isolated lineages may result in other evo-
torical changes in distribution (Hewitt 2000). lutionary consequences, such as reinforcement
Several organisms exhibit intraspecific genetic (Song et al. 2009, Lee & Mitchell-Olds 2011,
structures consistent with expected patterns from 2013), the genetic structure of once separate pop-
historical distributional changes suggested by pa- ulations in such contact zones are expected to
leontological studies (King & Ferris 1998, Abbott gradually dissolve through postglacial mixing as
et al. 2000, Palme & Vendramin 2002, Petit et al. long as gene flow between the lineages is not
NII-Electronic Library Service
The Japanese Society for Plant Systematics
2 Acta Phytotax. Geobot. Vol. 67
strongly restricted by reproductive isolation or species with different pollen dispersal modes in
other factors. the same area. Therefore, the effects of dispersal
In plants, the rate of genetic mixing may be mode on genetic structures remains unclear. We
related to differences in dispersability of pollen aimed to clarify the effects of pollen dispersal
and seeds. In particular, it is easy to expect that mode on the nrDNA genetic structure of several
species with different modes of pollen dispersal codistributed species that have different dispersal
would differ in dispersability. Based on data from modes but similar distribution ranges and similar
studies using allozyme markers, Hamrick et al. cpDNA genetic structures.
(1990) reported that wind-pollinated plant spe- Four deciduous broad-leaved tree species in
cies tend to show lower G values than animal- western Honshu, Japan, Carpinus laxiflora
sT
pollinated species. In other words, they suggested (Siebold & Zucc.) Blume, C. tschonoskii Maxim.,
that immigration rates and gene flow in wind-pol- C. japonica Blume, and Magnolia obovata
linated species are usually greater than in ani- Thunb., with various seed and/or pollen dispersal
mal-pollinated species. modes, have been reported to show similar east–
To examine the influence of different pollen west genetic differentiation patterns in their cpD-
dispersal modes on genetic structure, analyses NAs (Iwasaki et al. 2010, 2012). Furthermore,
using bi-parentally inherited nuclear genetic Tono et al. (2015) reported that the contact zones
markers are indispensable because maternally in- of the four species were highly consistent with
herited chloroplast DNA (cpDNA) markers can each other in terms of their cpDNAs. These ob-
migrate only through seeds, and thus are a reflec- servations indicate that the migration histories
tion of seed dispersal. In contrast, nuclear DNA and extent of postglacial genetic admixture
(nrDNA) markers migrate through both pollen through seeds were similar among the four spe-
and seeds. Thus, it cannot be determined whether cies. These species therefore provide us with a
pollen or seed dispersal is more influential when suitable system for examining the influence of
analyses are based only on nrDNA or cpDNA different pollen dispersal modes on genetic struc-
markers. ture.
Recently, several phylogeographic studies of In this study, we focused on two species, the
plants have used both nuclear and cytoplasmic insect-pollinated Magnolia obovata and the
DNA markers (Bai et al. 2010, Sakaguchi et al. wind-pollinated Carpinus laxiflora, which are a
2012, Qi et al. 2012, Ohtani et al. 2013, Bai et al. subset of the four species examined in the previ-
2014). For example, in wind-pollinated Juglans ous study based only on cpDNA markers (Tono et
mandshurica Maxim. and J. cathayensis Dode, al. 2015). To examine the influence of different
the genetic structure observed on the basis of pollen dispersal modes on genetic structure, we
nrDNA SSR markers was not as informative as compared the geographic patterns of genetic dif-
studies based on cpDNA markers (Bai et al. 2010, ferentiation based on nuclear microsatellite and
2014). In contrast, in insect-pollinated Kalopanax cpDNA markers between the insect-pollinated
septemlobus (Thunb.) Koidz. and Shorea lepro- M. obovata and the wind-pollinated C. laxiflora.
sula Miq., similar genetic structures were ob- We addressed the following two questions: (1)
served based on both nrDNA and cpDNA mark- Are the geographic patterns of genetic differen-
ers (Sakaguchi et al. 2012, Ohtani et al. 2013). tiation based on nrDNA markers similar to those
Those results suggest that differences in pollen based on cpDNA markers? (2) Are the geograph-
dispersability between wind- and insect-pollinat- ic patterns of genetic differentiation based on
ed species are significantly responsible for differ- nrDNA markers different between the two spe-
ences between cpDNA and nrDNA genetic struc- cies with different pollen dispersal modes?
tures.
The previous studies, however, did not com-
pare the genetic structure between codistributed
NII-Electronic Library Service
The Japanese Society for Plant Systematics
February 2016 Tono &al.– Postglacial lineage admixture in contact zones 3
molecular identifier (MID) tags. From the ob-
Materials and Methods tained sequences, candidates of microsatellite
loci were screened using MSATCOMMANDER
Sampling of the plants (Faircloth 2008). The sequences with either ≥8
We used silica gel to dry leaf samples from dinucleotide repeats, ≥8 trinucleotide repeats, ≥6
250 and 264 individuals of Magnolia obovata and tetra nucleotide repeats, ≥6 pentanucleotide re-
Carpinus laxiflora from 45 and 43 populations, peats, or ≥6 hexanucleotide repeats were searched
respectively, in the Kinki-Chugoku region (Ap- using Primer3 program (Rozen & Skaletsky
pendix 1 and Fig. 1). Of these, 224 and 232 sam- 2000) embedded in MSATCOMMANDER. For
ples from 42 and 38 populations, respectively, all loci, a 19-bp M13 tail (5’-CACGACGTTGTA-
had already been examined in a previous study AAACGAC-3’) was added to the 5’ end as the
(Tono et al. 2015). Thus, 26 and 32 individuals forward primer sequence following the method of
from 15 and 10 populations, respectively, were Schuelke (2000).
newly sampled in this study. Voucher specimens PCR amplification tests were performed for
have been deposited in the Makino Herbarium of eight individuals from six populations in a final
Tokyo Metropolitan University (MAK). volume of 5 μL using the standard protocol of the
QIAGEN Multiplex PCR Kit (QIAGEN) for sin-
cpDNA sequencing and haplotype classification gle-plex PCR (one primer pair per reaction). Each
Twenty-six samples of Magnolia obovata and reaction contained 0.2 μM reverse primer, 0.1 μM
32 samples of Carpinus laxiflora were newly ana- FAM-labeled M13 primer, and 0.1 μM forward
lyzed in this study. In addition, the cpDNA haplo- primer. PCR amplification included initial dena-
type data from the two species were cited by turation at 95°C for 15 min; 30 cycles of reactions
Tono et al. (2015). The methods for DNA extrac- at 94°C for 0.5 min, 57°C for 1.5 min, and 72°C
tion, polymerase chain reaction (PCR) amplifica- for 1 min; and a final extension at 60°C for 30
tion of cpDNA noncoding regions, sequencing, min. PCR products were analyzed using an ABI
and classification of the eastern and western hap- PRISM 3100 Genetic Analyzer (Applied Biosys-
lotypes followed those of Tono et al. (2015). The tems, Foster City, California, USA). PCR product
primers used for amplifying and sequencing the sizes were determined by comparison with Gen-
cpDNA non-coding regions are shown in Appen- eScan 500 or 600 LIZ Size Standard (Applied
dix 2. In total, cpDNA haplotypes were deter- Biosystems) using GeneScan analysis (Applied
mined for 250 and 264 individuals of M. obovata Biosystems). The results were analyzed using
and C. laxiflora from 45 and 43 populations, re- GeneMapper version 4.0 (Applied Biosystems).
spectively. The successfully amplified loci were ana-
lyzed for 31 individuals of Carpinus laxiflora
Development of microsatellite markers for Carpi- from one population (Gonami Pass, Ooi, Fukui
nus laxiflora Prefecture, Japan) with the same protocol used in
Microsatellite markers were developed the amplification test described above. For each
through pyrosequencing with enriched DNA li- locus, the number of alleles (N ), observed het-
A
braries. Total DNA was extracted from the leaf erozygosity (H ), expected heterozygosity (H ),
O E
samples of Carpinus laxiflora (voucher: AT3057) and fixation index (F ) were calculated using Ge-
IS
using the DNeasy Plant Mini Kit (QIAGEN, nAlEx version 6.5 (Peakall & Smouse 2006). The
Hilden, Germany). The DNA was shotgun se- Hardy-Weinberg equilibrium (HWE) was tested
quenced using the Roche 454 GS Junior using a for each locus using GenAlEx version 6.5 (Peak-
GS Junior Titanium Sequencing Kit (Roche Ap- all & Smouse 2006) with Bonferroni corrections.
plied Science, Penzberg, Germany) according to The linkage disequilibriums (LDs) between loci
the manufacturer’s instructions on quarter plates were examined using exact tests in FSTAT ver-
with libraries of three other species identified by sion 2.9.3.2 (Goudet 2002) with Bonferroni cor-
NII-Electronic Library Service
The Japanese Society for Plant Systematics
4 Acta Phytotax. Geobot. Vol. 67

FIG. 1. Sampling sites in the Kinki-Chugoku region. Detailed information for each site
Fig. 1. Sampling sites in the Kinki-Chugoku region. Detailed information for each site is shown in Appendix 1.
is shown in Appendix 1.

NII-Electronic Library Service
The Japanese Society for Plant Systematics
February 2016 Tono &al.– Postglacial lineage admixture in contact zones 5
rections. coefficient of genetic differentiation (F ) was ex-
st
amined to identify outlier loci with excessively
Microsatellite genotyping high or low F values compared with the neutral
st
For Magnolia obovata, seven microsatellite expectation under an island migration model.
markers developed by Isagi et al. (1999) (M10D6, This test was performed using an infinite allele
M6D3, M6D4, M15D5, M10D3, M6D10, and model with 50,000 simulations, and a confidence
M17D5) were used to determine the microsatel- interval of 0.95. LOSITAN analysis was per-
lite genotypes of 250 individuals from 45 popula- formed for each population that contained more
tions. The individuals were obtained from the than 15 individuals. Second, for Magnolia obo-
same samples that were used in the cpDNA anal- vata, we tested for departure from HWE at each
yses. For Carpinus laxiflora, 11 microsatellite locus for each population that contained more
markers developed in the present study (Carp2, than nine individuals using an exact test in Ge-
Carp3, Carp6, Carp11, Carp13, Carp16, Carp17, nAlEx version 6.5 (Peakall & Smouse 2006) with
Carp26, Carp27, Carp34, and Carp41) were used Bonferroni corrections. LD between loci was
to determine the microsatellite genotypes of 264 also tested using an exact test in FSTAT version
individuals from 43 populations. These individu- 2.9.3.2 (Goudet 2002). Bonferroni corrections
als were also obtained from the same samples were applied to all multiple statistical tests.
used in the cpDNA analyses. Primer information To evaluate genetic differentiation in the two
is shown in Appendix 3 and 4. For each primer species of trees, G (Nei 1987) was calculated
st
set, the forward primer was synthesized with a using FSTAT version 2.9.3.2 (Goudet 2002). In
tag sequence (FAM = 5’-CACGACGTTGTA- addition, G’ and standardized values of G
st st
AAACGAC-3’; NED = 5’-CTATAGGGCACGC- were calculated manually (Hedrick 2005). To elu-
GTGGT-3’ VIC = 5’-TGTGGAATTGTGAGC- cidate the characteristics of each population that
GG-3’) added to its 5’ end using the method of contained more than four individuals, the follow-
Schuelke (2000). PCR amplification was per- ing parameters were calculated on the basis of the
formed with a final volume of 5 μL using the multilocus genotype data using GenAlEx version
standard protocol of the QIAGEN Multiplex PCR 6.5 (Peakall & Smouse 2006) and FSTAT version
Kit (QIAGEN) with two- or three-plex PCR (two 2.9.3.2: allelic richness (AR), the total number of
or three primer pairs per reaction). Three primers private alleles (PA), observed heterozygosity
were used in each reaction: 0.066 μM reverse (H ), and expected heterozygosity (H ). The val-
o e
primer, 0.033 μM fluorescently labeled primer, ues for AR and PA were calculated by rarefying to
and 0.033 μM forward primer. The methods for five individuals using FSTAT version 2.9.3.2 and
PCR amplification and determination of the prod- HP-RARE version 1.1 (Kalinowski 2005), re-
uct size were the same as those for microsatellite spectively.
marker development, except for different anneal-
ing temperatures (50–57°C). STRUCTURE analysis
The intraspecific genetic structure of the two
Microsatellite data analyses species was estimated by Bayesian clustering us-
To check whether each locus met the require- ing STRUCTURE version 2.3.4 (Pritchard et al.
ments for neutral evolution, population genetic 2000, Falush et al. 2003, 2007). This analysis can
analysis was performed. At first, neutral evolu- reveal the population structure with a priori as-
tion of the analyzed microsatellite markers was signment of individuals to populations on the ba-
tested for both species using the F -outlier ap- sis of multilocus genotype data. The assignment
st
proach (Beaumont & Nichols 1996, Beaumont probabilities of clusters were estimated using the
2005) implemented in LOSITAN (Antao et al. Markov Chain Monte Carlo (MCMC) method,
2008). In this analysis, the relationship between assuming gene pools with the least possible LD
the expected heterozygosity (H ) and Wright’s and smallest possible departure from HWE. The
e
NII-Electronic Library Service
The Japanese Society for Plant Systematics
6 Acta Phytotax. Geobot. Vol. 67
admixture model and LOCPRIOR model with Pollen/seed migration ratios
correlated allele frequencies were used (Falush et To estimate the relative rates of pollen and
al. 2003, Hubisz et al. 2009). Ten independent seed migration among populations, the pollen/
runs were performed using a burn-in period of seed migration ratio (r) was calculated using the
100,000 and posterior probabilities were obtained following equation: (pollen flow/seed flow) =
from 100,000 iterations for each number of clus- {(1/G’ – 1) – 2 (1/G’ – 1)}/(1/G’ – 1), where
STn STc STc
ters, K = 1 to 10. The most likely value of K was G’ is the G’ of cpDNA and G’ is the G’ of
STc ST STn ST
assessed using the ΔK values (Evanno et al. nrDNA. This is a modification of the equation by
2005). Ennos (1994) with the substitution of G values
ST

Fig. 2 . (A) Geographic distribution pattern of two clusters estimated by STRUCTURE analysis on basis of nuclear microsatel-
lite data and composition of chloroplast DNA haplotypes for Magnolia obovata (a) and Carpinus laxiflora (b). For nuclear
m icrosatellite data, blue represents Cluster I; red represents Cluster II. For chloroplast DNA, eastern and western haplo-
types are in blue and red, respectively. Number of samples analyzed per population and haplotype composition is propor-
t ional to circle size. Population numbers correspond to those in Appendix 1.

NII-Electronic Library Service
The Japanese Society for Plant Systematics

February 2016 Tono &al.– Postglacial lineage admixture in contact zones 7
for F values. The values of G and G’ (stan- the eastern and western parts of the Kinki-Chu-
ST ST ST
dardized values of G ; Hedrick 2005) of cpDNA goku region. The general tendencies were the
ST
were calculated by PERMUT version 2.0 (http:// same as in previous studies (Iwasaki et al. 2012,
www.pierroton.inra.fr/genetics/labo/Software/) Tono et al. 2015).
(Pons & Petit 1996) and manually, respectively.

Microsatellite marker development for Carpinus
Results laxiflora
The de novo pyrosequencing produced 15,602

Geographic distribution patterns of cpDNA hap- reads with an average length of 435.62 bp. Three
lotypes hundred and sixty three sequences with simple
In Magnolia obovata, five haplotypes (A, B, sequence repeats were found by MSATCOM-

C, F, and G) were found. Three (A, C, and F) and MANDER. Primer sets were successfully de-
two haplotypes (B and G) were classified as east- signed for a total of 41 microsatellite loci. Of
ern and western haplotypes, respectively, accord- these, 14 primer pairs (Carp2, Carp3, Carp6,
ing to Tono et al. (2015). In Carpinus laxiflora, Carp11, Carp13, Carp16, Carp17, Carp20, Carp21
two haplotypes (A and B) were found. Haplotype Carp24, Carp26, Carp27, Carp34, and Carp41)

B and A were classified as eastern and western successfully amplified DNA fragments through
haplotypes, respectively, according to Tono et al. PCR (Appendix 4). The remaining 27 primer
(2015). All of the haplotypes observed in the two pairs failed to amplify DNA fragments with the

species had already been reported in Tono et al. expected product size. Characteristics of these 14
(2015). The geographic distribution patterns of microsatellite markers are shown in Appendix 5.
the eastern and western cpDNA haplotypes in the Only one marker (Carp21) showed a significant

two species are shown in Fig. 2. After combining deviation from HWE for the test population (P <
the newly obtained data from this study with data 0.05 after Bonferroni correction). Significant LD
cited in Tono et al. (2015), proportions of the east- was not detected for any pairs of loci. It was dif-

ern and western haplotypes still differed between ficult to score fragment sizes for two markers

Fig. 2. (B) Assignment probabilities into two STRUCTURE clusters (top) and cpDNA haplotypes (bottom) for each individu-
al. Blue represents proportion of cluster I; red represents cluster II. In chloroplast DNA results, blue represents eastern
haplotypes; red represents western haplotypes. Numbers at bottom represent population numbers. Populations are sorted
from east to west.
NII-Electronic Library Service
The Japanese Society for Plant Systematics
8 Acta Phytotax. Geobot. Vol. 67
tAble 1. Parameters of genetic differentiation of chloroplasts, nuclear DNA polymorphisms, and pollen/seed migration ratios.
Magnolia obovata Carpinus laxiflora
Chloroplasts G (G ) 0.400 0.675
ST STc
Chloroplasts G’ (G’ ) 0.675 0.827
ST STc
Nuclear G (G ) 0.044 0.029
ST STn
Nuclear G’ (G’ ) 0.281 0.124
ST STn
Pollen /seed migration ratio r 3.331 31.608
(Carp20 and Carp24) from 264 individuals in 43 had higher G' values than C. laxiflora.
ST
populations . Therefore, these three markers (Carp20,
Carp21, and Carp24) were excluded from further Genetic structure estimated using nuclear micro-
genetic analyses. Consequently, 11 nuclear mic- satellite markers
rosatellite markers were used for genetic analy- In both species, the value for ΔK was the high-
ses. est when K = 2 (Appendix 8). Therefore, we con-
sidered the appropriate number of clusters (K) to
Characteristics of the microsatellite loci of Mag- be two. The assignment probabilities into two
nolia obovata and Carpinus laxiflora clusters for each individual are shown in Fig. 4
In Magnolia obovata, only one marker with information from the cpDNA haplotypes
(M6D10) showed significant deviation from (Tono et al. 2015). Possible admixing individuals
HWE in one (No. 40) of the eight populations ex- with nearly the same levels of probabilities of
amined (P < 0.05 after Bonferroni correction). cluster I and II were observed in several popula-
Significant LD was not detected for any pairs of tions of the two species.
loci in any populations. No outlier loci were de- Geographic distributions of the STRUC-
tected with excessively high or low F values TURE clusters in the two species are shown in
ST
compared with the neutral expectation. Based on Fig. 2, together with those of the cpDNA haplo-
these results, all seven loci were used for further types. In Magnolia obovata, the proportion of
analyses. cluster I (in blue) was high in populations from
In Magnolia obovata, AR ranged from 3.94 to the eastern and Sea of Japan side of the Kinki-
5.5, PA from 0 to 0.24, H from 0.667 to 0.833, Chugoku region, whereas the proportion of clus-
O
and H from 0.7 to 0.84 (Appendix 6). The mean ter II (in red) was high in the populations from
E
values of AR, PA, H , and H were 4.87, 0.10, 0.71, the western and Pacific Ocean side of the region.
O E
and 0.71, respectively. In C. laxiflora, the proportion of cluster I (in blue)
In Carpinus laxiflora, AR ranged from 3.94 to was high in populations from the eastern part of
5.5, PA from 0 to 0.31, H from 0.6 to 0.795, and the Kinki-Chugoku region, whereas the propor-
O
H from 0.634 to 0.773 (Appendix 7). The mean tion of cluster II (in red) was high in populations
E
values of AR, PA, H , and H were 5.59, 0.18, 0.76, from the western part of the region.
O E
and 0.78, respectively. No outlier loci were de-
tected with excessively high or low F values Pollen/seed migration ratio
ST
compared with the neutral expectation. The overall genetic differentiation among
The overall genetic differentiation among populations (G’ ) at cpDNA of Magnolia obo-
ST
populations at the 7 and 11 loci for Magnolia obo- vata and Carpinus laxiflora was 0.675 and 0.827,
vata and Carpinus laxiflora, respectively, was respectively. The pollen/seed migration ratio r of
low (G , G' = 0.044, 0.281 and 0.029, 0.124 M. obovata and C. laxiflora estimated on the ba-
ST ST
among the M. obovata and C. laxiflora popula- sis of the values of G' was 3.33 and 31.60, re-
ST
tions, respectively) (Table 1). Magnolia obovata spectively (Table 1).
NII-Electronic Library Service
The Japanese Society for Plant Systematics
February 2016 Tono &al.– Postglacial lineage admixture in contact zones 9
Carpinus laxiflora (r = 31.6) was much higher
Discussion than in Mgnolia obovata (r = 3.33) (Table 1). At
the same time, the geographic distribution pat-
Two nuclear genetic clusters were detected in terns of cpDNA haplotypes in the two species re-
both species by STRUCTURE analysis. In Mag- ported by Tono et al. (2015) were very similar,
nolia obovata, the results based on nuclear micro- suggesting that migration rates through seeds dif-
satellites and cpDNA haplotypes were highly fer little between the two species. The large dif-
consistent (Fig. 2). In contrast, in Carpinus laxi- ferences in nuclear genetic structure as well as in
flora, the genetic structure based on nuclear mic- the values of the pollen/seed migration ratios be-
rosatellites and cpDNA haplotypes was not as tween the two species likely reflect differences in
concordant as in M. obovata (Fig. 2). Moreover, the amount of gene flow through pollen.
in M. obovata, the difference in frequencies in Dissolution of the genetic structure after the
clusters 1 and 2 was observed not only between LGM may be related to differences in dispers-
the eastern and western parts but also between ability through pollen and seeds. Hamrick et al.
the northern (Sea of Japan side) and southern (1990) indicated that wind-pollinated species
parts of the Kinki-Chugoku region. In previous have a higher rate of gene flow than do animal-
phylogeographic studies of M. obovata (Iwasaki pollinated species based on differences in G
ST
et al. 2012), three major cpDNA haplotypes with values between wind- and animal-pollinated spe-
different main distribution ranges were found: cies. For example, wind-pollinated Fagus crena-
haplotype A in eastern Japan, B in southwestern ta Blume was reported to show marked genetic
Japan, and C in the Sea of Japan side of Honshu. structure in a molecular phylogeographic analy-
They concluded that the genetic structure could sis based on cpDNA markers (Fujii et al. 2002).
have been shaped by isolation of the populations However, the genetic structure based on nrDNA
into several different refugia in each of the areas SSR markers was not as clear as studies based on
during the last glacial period. In this study, indi- cpDNA markers (Hiraoka & Tomaru 2009). Sim-
viduals with the nuclear cluster I (blue) and cpD- ilarly, in wind-pollinated Betula maximowic-
NA haplotype C and those with the nuclear clus- ziana Regel, Picea jezoensis Maxim., Cercidi-
ter II (red) and cpDNA haplotype B were fre- phyllum japonicum Siebold & Zucc., Juglans
quently observed in the Sea of Japan side area and mandshurica, and J. cathayensis, the genetic
western Japan, respectively. Therefore, the north- structures observed on the basis of nrDNA SSR
east-southwest genetic divergence of the nrDNA markers were also not as clear as those based on
observed in the Kinki-Chugoku region in this cpDNA markers (Tsuda & Ide 2005, 2010, Aiza-
study may be mainly attributed to historical iso- wa et al. 2007, 2009, Bai et al. 2010, 2014, Qi et
lation of the populations between the Sea of Japan al., 2012). In contrast, insect-pollinated Melam-
side refugia and the southwestern Japan refugia podium leucanthum Torr. & A. Gray, Sinopodo-
during LGM. However, further studies with wid- phyllum hexandrum (Royle) T. S. Ying, Cyanan-
er sampling are necessary to reveal the migration thus delavayi Franch., Kalopanax septemlobus,
history that shaped the northeast-southwest dif- Shorea leprosula, and Callicarpa japonica
ferentiation in this region. Thunb., showed similar and marked genetic
Maternally inherited cpDNA can only be dis- structure based on both nrDNA and cpDNA
persed through seeds, whereas bi-parentally in- markers (Rebernig et al. 2010, Li et al. 2011,
herited nrDNA can be dispersed through seeds 2012, Sakaguchi et al. 2012, Ohtani et al. 2013,
and pollen. Therefore, if geographical distribu- Hirano et al. 2014).
tion patterns for nrDNA clusters and cpDNA hap- The results concordantly suggest that differ-
lotypes are consistent, it means that dispersabili- ences between the genetic structures revealed by
ty through seeds and through pollen differs little. cpDNA and nrDNA markers may be caused by
Additionally, the pollen/seed migration ratio in the movement of pollen, which is greater than
NII-Electronic Library Service
The Japanese Society for Plant Systematics
10 Acta Phytotax. Geobot. Vol. 67
movement of seeds in wind-pollinated species. their valuable advice. This study was partly supported by
The wind-pollinated Carpinus laxiflora plausibly the Research Project “A new cultural and historical explo-
ration into human-nature relationships in the Japanese ar-
has a higher migration rate through pollens dis-
chipelago” of the Research Institute for Humanity and
persal than in the insect-pollinated Magnolia ob-
Nature to N.M and A.S.
ovata. Accordingly, differences in nrDNA genet-
ic structure between M. obovata and C. laxiflora
is likely due to difference in the mode of dispersal References
of pollen.
Results from this study help to clarify the in- Abbott, R. J., L. C. Smith, R. I. Milne, R. M. M. Craw-
ford, K. Wolff & J. Balfour. 2000. Molecular analysis
fluence of life history traits, such as mode of pol-
of plant migration and refugia in the Arctic. Science
len dispersal on dissolution of genetic structure.
289: 1343–1346.
Hamrick et al. (1990) reported that the mode of
Aizawa, M., H. Yoshimaru, H. Saito, T. Katsuki, T.
pollen dispersal has a large influence on the ge- Kawahara, K. Kitamura, F. Shi & M. Kaji. 2007. Phy-
netic structure of plants. By considering phyloge- logeography of a northeast Asian spruce, Picea
jezoensis, inferred from genetic variation observed in
netic constraints, Duminil et al. (2007) suggested
organelle DNA markers. Molec. Ecol. 16: 3393–3405.
that differences in the mode of pollen dispersal
Aizawa, M., H. Yoshimaru, H. Saito, T. Katsuki, T.
do not affect genetic differentiation patterns.
Kawahara, K. Kitamura, F. Shi, R. Sabirov & M.
These suggestions by two previous studies are Kaji. 2009. Range-wide genetic structure in a north-
important but contradictory. Several reasons can east Asian spruce (Picea jezoensis) determined using
nuclear microsatellite markers. J. Biogeogr. 36: 996–
be postulated for this contradiction. Duminil et
1007.
al. (2007) did not consider the effects of differ-
Antao, T., A. Lopes, R. J. Lopes, A. Beja-Pereira & G.
ences in the migration histories in their meta-
Luikart. 2008. LOSITAN: A workbench to detect mo-
analyses, although many phylogeographic studies lecular adaptation base on a F -outlier method. BMC
ST
have reported different spatial genetic structures Bioinforma 9: 323.
Bai, W. N., W. J. Liao & D. Y. Zhang. 2010. Nuclear and
and different migration histories for various spe-
chloroplast DNA phylogeography reveal two refuge
cies. Furthermore, their studies have included
areas with asymmetrical gene flow in a temperate
various geographic study systems whose topo-
walnut tree from East Asia. New Phytol. 188: 892–
graphic factors and geographic scales are differ- 901.
ent. Such differences may influence migration Bai, W. N., W. T. Wang & D. Y. Zhang. 2014. Contrasts
between the phylogeographic patterns of chloroplast
histories and patterns of genetic differentiation.
and nuclear DNA highlight a role for pollen-mediated
Thus, comparative studies focused on co-distrib-
gene flow in preventing population divergence in an
uted species with similar migration histories in
East Asian temperate tree. Molec. Phylogen. Evol. 81:
the same geographic research system, such as 37–48.
contact zones as in the present study, will provide Beaumont, M. A. & R. A. Nichols. 1996. Evaluating loci
for use in the genetic analysis of population structure.
valuable suggestions about relationships between
Proc. Roy. Soc. London, Ser. B, Biol. Sci. 263: 1619–
life historical traits and population genetic struc-
1626.
ture. This study demonstrates that contact zones
Beaumont, M. A. 2005. Adaptation and speciation: what
can be good systems for examining the effects of can FST tell us? Trends Ecol. Evol. 20: 435–440.
life history traits, such as pollen dispersability, on Duminil, J., S. Fineschi, A. Hampe, P. Jordano, D. Salvi-
ni, G. G. Vendramin & R. J. Petit. 2007. Can popula-
genetic structure.
tion genetic structure be predicted from Life-History
traits? Amer. Naturalist 169: 662–672.
We thank Mr. R. Nitta (Makino Herbarium, Tokyo Met-
Ennos, R. A. 1994. Estimating the relative rates of pollen
ropolitan University), Dr. K. Sugai (Department of Forest
and seed migration among plant populations. Hered-
Genetics, Forestry and Products Research Institute), and
ity 72: 250–259.
Dr. E. Oguri (Department of Biological Science, Hiroshi-
Evanno, G., S. Regnaut & J. Goudet. 2005. Detecting the
ma University) for their technical support. We also thank
number of clusters of individuals using the software
Dr. T. Sugawara, Dr. H. Kato, and Dr. Y. Kakugawa
STRUCTURE: a simulation study. Molec. Ecol. 14:
(Makino Herbarium, Tokyo Metropolitan University) for
2611–2620.
NII-Electronic Library Service