Geographic variation of chloroplast DNA in Platycarya strobilacea (Juglandaceae)


  • Shi-Chao CHEN,

    Corresponding author
    1. ( College of Life Science and Technology, Tongji University, Shanghai 200092, China)
    2. ( Tongji University-Lishui Institute of Traditional Chinese Medicine, Lishui 323000, China)
       Authors for correspondence. S.-C. Chen. E-mail:; Tel.: 86-21-65982587; Fax: 86-21-65981041. C.-X. Fu. E-mail:; Tel.: 86-571-88206607; Fax: 86-571-86432273.
    Search for more papers by this author
  • Li ZHANG,

    1. ( College of Life Science and Technology, Tongji University, Shanghai 200092, China)
    Search for more papers by this author
  • Jie ZENG,

    1. ( College of Life Science and Technology, Tongji University, Shanghai 200092, China)
    Search for more papers by this author
  • Fei SHI,

    1. ( College of Life Science and Technology, Tongji University, Shanghai 200092, China)
    Search for more papers by this author
  • Hong YANG,

    1. ( School of Medicine, Tongji University, Shanghai 200092, China)
    Search for more papers by this author
  • Yun-Rui MAO,

    1. ( Laboratory of Systematic and Evolutionary Botany, State Conservation Center for Gene Resources of Endangered Wildlife, College of Life Sciences, Zhejiang University, Hangzhou 310058, China)
    Search for more papers by this author
  • Cheng-Xin FU

    Corresponding author
    1. ( Laboratory of Systematic and Evolutionary Botany, State Conservation Center for Gene Resources of Endangered Wildlife, College of Life Sciences, Zhejiang University, Hangzhou 310058, China)
       Authors for correspondence. S.-C. Chen. E-mail:; Tel.: 86-21-65982587; Fax: 86-21-65981041. C.-X. Fu. E-mail:; Tel.: 86-571-88206607; Fax: 86-571-86432273.
    Search for more papers by this author

 Authors for correspondence. S.-C. Chen. E-mail:; Tel.: 86-21-65982587; Fax: 86-21-65981041. C.-X. Fu. E-mail:; Tel.: 86-571-88206607; Fax: 86-571-86432273.


Abstract  The monotypic genus Platycarya (Juglandaceae) is one of the most widespread temperate tree species in East Asia. In this research, we implemented a phylogeographical study using chloroplast DNA (cpDNA) (psbA-trnH and atpB-rbcL intergenic spacer) sequences on Platycarya strobilacea, in order to identify the locations of the species’ main refugia and migration routes. A total of 180 individuals of P. stobilacea from 27 populations from China and Jeju Island (Korea) were collected. The results revealed that P. strobilacea had 35 haplotypes for the two intergenic spacers and high genetic diversity (hT= 0.926). This surprisingly high diversity of haplotypes indicates its long evolutionary history, which is in agreement with previous phylogenetic analyses and fossil records. Significant cpDNA population subdivision was detected (GST= 0.720; NST= 0.862), suggesting low levels of recurrent gene flow through seeds among populations and significant phylogeographical structure (NST > GST, P < 0.05). The construction of phylogenetic relationships of the 35 chlorotypes detected four major cpDNA clades. Divergence dating analyses using BEAST suggest that the divergence of the major cpDNA clades occurred before the Miocene. Demographic analysis indicated that the Eastern clade underwent localized demographic expansions. The molecular phylogenetic data, together with the geographic distribution of the haplotypes, suggest the existence of multiple glacial refugia in most of its current range in China through Quaternary climatic oscillations.

The Sino-Japanese Floristic Region (SJFR) of East Asia harbors the world's most diverse temperate flora. However, plant phylogeographic studies in this region are sparse compared to other temperate regions (e.g., Europe and North American), especially in terms of large-scale sampling with sufficient coverage in this region. The Fagales order is a prominent component of extant tropical, subtropical and temperate forests in the world, whose current geographic distribution indicates the influence of both past and present ecological factors (Avise, 1992; Fujii et al., 1997; Soltis et al., 1997; Aoki et al., 2004b; Wu et al., 2007). However, relatively few phylogeographic studies have been conducted on this important group in China (Wang et al., 2009; Qiu et al., 2011), and most previous studies on Fagus (Tomaru et al., 1997; Fujii et al., 2002; Okaura & Harada, 2002; Hiraoka & Tomaru, 2009a, 2009b) and Quercus (Kanno et al., 2004; Okaura et al., 2007), have been restricted to the Japanese archipelago. The monotypic genus Platycarya (Juglandaceae) of Fagales, that is distributed almost throughout the entire East Asian temperate forests (Lu, 1982), is of particular interest to evolutionary biologists and biogeographers.

Platycarya strobilacea, commonly known as species endemic to East Asia, is a small deciduous tree (Kuang & Lu, 1979; Zhou & Momohara, 2005). It is native to China, Korea, Japan and Vietnam, and is known for some ancient characters, such as bisexual inflorescence aggregated on apices of branches, bracts separated with ovary and fruits, and the first pair of true leaves of the sprout being a simple leaf (Hjelmqvist, 1951; Manning, 1978; Manos & Stone, 2001; Li et al., 2005). Platycarya occupies a unique systematic position in Juglandaceae, and it is a sister group to Juglandeae based on molecular evidence (Manos & Stone, 2001; Manos et al., 2007; Xiang, 2010). Fossil records indicate that Platycarya was widely distributed in all continents of the Northern Hemisphere during the early stage of the Tertiary period (i.e. Paleocene to Eocene); however, during the Quaternary ice age it became extinct in many places, but survived in East Asia (Lu, 1982; Zhou & Momohara, 2005). Although many studies have investigated its agricultural, nutritional, and medical values (Li et al., 2005), little is known about its phylogeographic patterns and population genetics.

The chloroplast genome in Platycarya is maternally inherited, as it is in other Juglandaceae (Xiang, 2010), and thus is only transmitted through seeds. Consequently, the geographical variation of cpDNA in this species is expected to reflect patterns of historical seed flow and colonization. In the present paper, using cpDNA haplotype sequences, we assessed patterns of geographical molecular variation in most of all presently known populations of P. strobilacea. The specific aims are to identify the locations of the species’ main refugia and infer the demographic history that has led to the contemporary spatial genetic structure of chloroplast variation.

1 Materials and methods

1.1 Population sampling

A total of 180 individuals of P. stobilacea from 27 populations, comprising 1–16 individuals per population, were sampled over its range in China (except Taiwan), and one population from Jeju Island, Korea (Table 1). Within each population, foliar samples were collected from individuals as far apart as possible (typically >10 m) in order to avoid samples being collected from close relatives. The fresh leaves were dried in silica gel, and most were stored in a freezer (−20 °C) but some were stored at room temperature. A sample of Juglans regia from Tianmu Mountain in Zhejiang Provence was also collected as an outgroup. Voucher specimens were deposited in the herbarium of Tongji University.

Table 1.  Localities and numbers of samples surveyed across the natural range of Platycarya strobilacea and the numbers of haplotypes observed in the populations surveyed
PopulationLatitude (N)Longitude (E)Altitude (m) n s n h h π
  1. n s, the number of samples analysed; nh, the number of haplotypes observed; h, haplotypic diversity index.

Yaoluoping, Anhui, China (AH)30°59′2.66″116°08′14.83″11141810.0000.000 00
Baibao, Wenshan, Yunnan, China (BB)23°45′25.1″105°30′9.4″ 955730.5240.000 77
Tianlin, Baise, Guangxi, China (BS)24°16′40.1″106°13′20.1″ 649520.4000.000 72
Jingfe Mt., Chongqin, China (CQ)29°02′36.59″107°11′21.23″1489820.2500.000 67
Longtan, Fujian, China (FJ)25°52′23″119°32′25″ 300510.0000.000 00
Zhaoqin, Guangdong, China (GD)23°04′45.5″112°28′23.3″ 114110.0000.000 00
Xishan Mt., Guilin, Guangxi, China (GL)25°23′23.6″110°02′16″ 301520.4000.000 72
Wenxian, Gansu, China (GS)32°46′19.2″105°21′17.1″ 650830.4640.000 84
Diecai Mt., Guilin, Guangxi, China (GX)25°17′28.91″110°17′49.85″ 1671410.0000.000 00
Tuchen, Yichang, Hubei, China (HB)30°39′19.9″111°06′21.1″ 145710.0000.000 00
Songshan Mt., Henan, China (HN)34°28′34.7″112°56′21.3″1135810.0000.000 00
Huangshan Mt., Anhui, China (HS)30°07′54.16″118°07′44.25″ 709410.0000.000 00
Lushan Mt., Jiangxi, China (JX)29°34′21.2″115°50′27″1100331.0000.006 59
Xishan Mt., Kunming, Yunnan, China (KM)25°04′4.8″102°42′15.5″1933611.0000.001 80
Jeju Island, Korea (KO)33° 23′15.0″126°48′39.0″ 148220.0000.000 00
Liupanshui, Guizhou, China (LP)26°36′00.32″104°51′12.74″1932320.6670.001 20
Baiyun Mt., Lishui, Zhejiang, China (LS)28°30′41.48″119°54′46.58″ 446410.0000.000 00
Ningxiang, Hunan, China (NX)28°08′15.3″112°01′26.5″ 244221.0000.002 70
Pingshi, Lechang, Guangdong, China (PS)25°17′20″113°03′22.1″ 228720.2860.001 28
Qingyuan, Lishui, Zhejiang, China (QY)27°43′02.82″119°11′06.13″ 1434810.0000.000 00
Laoshan Mt., Shandong, China (SD)36°06′37.2″120°29′45.6″ 143620.3330.000 90
Zhengpin, Shanxi, China (SX)31°53′47.2″109°31′21.7″1402620.5330.001 44
Lingkou, Tongdao, Hunan, China (TD)26°16′18.8″109°50′30.2″ 499430.8330.001 50
Tianmu Mt., Zhejiang, China (TM)30°12′47.1″119°23′12.4″ 4351610.0000.000 00
Wudan Mt., Hubei, China (WD)32°24’38.54″111°1’24.04″ 770410.0000.000 00
Xinyi, Guizhou, China (XY)25°08′4.1″104°57′16.4″10121360.8460.001 68
Zhunyi, Guizhou, China (ZY)27°56′4.7″106°44′45.4″1225620.3330.000 60
Total   180350.9260.005 82

1.2 Sequencing of noncoding regions of cpDNA

Total DNA was extracted using a 2×CTAB (hexadecyl trimethyl ammonium bromide) buffer according to the method of Doyle & Doyle (1987). The PCR amplification was conducted in a total reaction volume of 15 μL containing 10–20 ng of total DNA, each primer at 0.15 pmol/L, 0.1 mmol/L deoxynucleoside triphosphates (dNTPs), 50 mmol/L KCl, 2 mmol/L MgCl2, 10 mmol/L Tris-HCl (pH 8.3) and 0.375 units of Taq DNA polymerase (Shanghai Bocai Biotech Company, Shanghai, China). Double-stranded DNA was amplified after incubation at 94 °C for 3 min, followed by 30 cycles of incubation at 94 °C for 30 s, 55 °C for 30 s and 72 °C for 30 s, with a final extension at 72 °C for 15 min. To examine the geographical distribution of cpDNA in p. stobilacea, we amplified two noncoding regions of cpDNA, atpB-rbcL and psbA-trnH intergenic spacer regions by using the primers reported by Terachi (1993) and Kress & Erickson (2007), which had been successfully amplified and exhibited some variations in our preliminary experiments. One pair of primers was newly designed and more effective for the atpB-rbcL region (F: TTCTCGCAACAACAAGGTCT; R: AAACCCCAGGACCAGAAGTA). Primers used for PCR amplification were also used as sequencing primers. The PCR products were checked using electrophoresis in 1.0% agarose gels, and then used as templates for direct sequencing. Sequencing was conducted from both ends using the above primers. For some minor variants, we performed the sequencing at least twice.

1.3 Phylogenetic analyses

Sequences were checked and aligned using Geneious Pro v5.5 (Drummond et al., 2011). We employed simple indel coding for gap coding (Simmons & Ochoterena, 2000). Multinucleotide units and multinucleotide repeat units were coded as binary states 0 or 1 in the phylogenetic analysis. The gaps caused by the mononucleotide repeat units were removed in phylogenetic analysis, because homology is highly uncertain for these repeated nucleotides (Kelchner, 2000). Phylogenetic analyses were performed using maximum parsimony (MP) and maximum likelihood (ML) analyses on the combined data. The MP analyses were carried out using PAUP* version 4.0 b10 (Swofford, 2002) through a heuristic search with tree-bisection-reconnection (TBR) branch swapping and the MulTrees option. All characters were equally weighted in the analyses. A strict consensus tree of MP trees was reconstructed, followed by DELTRAN character optimization. To estimate the confidence levels of monophyletic groups, we performed bootstrap analyses for 1000 replicates with the same tree search procedure as described. The ML analyses were conducted using PAUP* version 4.0b10, applying the best-fit molecular evolution model and parameters determined using Modeltest version 3.7 (Posada & Crandall, 1998) with the Akaike information criterion (AIC). The best-fit models for cpDNA were GTR. Based on the parameter values suggested by Modeltest, a heuristic search using the neighbor joining (NJ) tree as the starting tree was run with PAUP* version 4.0b10. Clade supports were assessed by nonparametric bootstrapping using 1000 replicates with a heuristic search strategy.

The haplotype network was constructed using TCS version 1.21 (Clement et al., 2000) with 95% confidence limits and the ‘gaps missing’ option turned on. The same sequence alignment assembled for the MP and ML analysis was used for the network analysis, but the gaps coded as binary states ‘0 or 1’ were coded again as binary states ‘A or T’ for the TCS.

1.4 Molecular dating

A Bayesian analysis of combined data was also used to estimate the divergence times of the major lineages to the most recent common ancestor (TMRCA) by software BEAST version 1.5.3 (Drummond & Rambaut, 2007). Based on a comparison of eight chloroplast genes for angiosperms (Wolfe et al., 1987), a rate of about 1.0–3.0 × 10−9 substitutions per neutral site per year (s/s/y) was used to obtain absolute values of TMRCA. All of the analyses were performed using the GTR model of nucleotide substitution.

1.5 Population genetic diversity and genetic differentiation

Basic sequence statistics including haplotype diversity (H) and nucleotide diversity (π) were calculated using the DnaSP v5.10 program (Librado & Rozas, 2009). Three analyses of molecular variance (AMOVA) were performed to determine how the genetic variation is distributed within and among collection sites along the resulting genetic groups. In all analyses, 1000 permutations were run to obtain test statistics using ARLEQUIN 3.5 (Excoffier & Lischer, 2010). To infer other possible groups without a prior user defined structure, we performed a spatial analysis of molecular variance (SAMOVA) with the computer software SAVMOA v1 (Dupanloup et al., 2002). As an overall assessment of geographical structure affecting the population differentiation, a comparison of the two fixation indices, GST and NST, was carried out using the DnaSP 5.10 program (Librado & Rozas, 2009).

1.6 Demographic history

In order to test for evidence of demographic range expansions, we examined the distribution of pairwise frequencies within each group (mismatch distributions). In addition, we calculated the Tajima's D statistic (Takahata et al., 1989) to test demographic range expansion. Significant values of D can be attributed to bottlenecks, selective effects, population expansion or heterogeneity of mutation rates. Finally, we used Fu's Fs test, which uses haplotype distribution information to test for demographic expansion and is more sensitive to population growth than Tajima's D. Demographic analyses were carried out for the resulting groups using the DnaSP v5.10 program (Librado & Rozas, 2009).

2 Results

2.1 Intraspecific cpDNA variation, diversities and genetic differentiation

Within the 181 P. strobilacea samples from 26 sites, the nucleotide sequence lengths were 310–318 bp and 921–926 bp for the psbA-trnH and atpB-rbcL regions, respectively. A partition-homogeneity test of the two chloroplast fragments yielded non-significant results (P= 0.4); since the chloroplast is inherited as a single unit, the results we report herein are for the two fragments combined. The length after multiple alignments of the combined sequences was 1112 bp, and a total of 35 haplotypes were recognized from P. strobilacea. Polymorphic sites and sequence variability of the aligned sequences are summarized in Table 2.

Table 2.  Variable nucleotide sites and length polymorphisms of cpDNA (atpB-rbcL and psbA-trnH) sequences in the Platycarya strobilacea, identifying 36 haplotypes (Hap1–Hap36)
PopulationHaplo- type atpB-rbcL
BB, BS, XYHap2..0a..........C..1e................G.C...
BB, BS, XYHap3..0a.A.0c......C..1e................G.C...
CQ, ZYHap5..0a.......A..C..1e................G.C...
FJ, HS, QY, SD, TMHap7.......1d.......................0l.G.C...
GL, GSHap9...................1f.............G.C...
GS, HNHap12............G..C.......................
GS, SXHap13............G..C.......................
HB, NXHap14............G..C.......................
WD, KOHap17...................1f1g..................
PopulationHaplo- type psbA-trnH
  1. All sequences are compared to the reference Hap1. Number 1/0 in sequences denote presence/absence of length polymorphism, identified by superscript letters (a–t). Note that ploynuleotide stretches (poly A) were excluded from the analysis. Population codes are identified in Table 1.


BB, BS, XYHap2......G....C......
BB, BS, XYHap3......G....C......
CQ, ZYHap5......G....C......
FJ, HS, QY, SD, TMHap7...........C....C.
GL, GSHap9...........C......
GS, HNHap12....A..A...C......
GS, SXHap13.......AG..C......
HB, NXHap14.......A...C......
WD, KOHap17..................

The sample size, number of haplotypes, values of nucleotide diversity (π), and haplotype diversity (h) within each population are presented in Table 1. At the species wide scale, the cpDNA data revealed high estimates of haplotype diversity (hT= 0.926), and nucleotide diversity (πT= 0.0058). However, the π value varies greatly among populations from 0 to 0.006 59. The SAMOVA revealed increasing FCT values when the number of groups increased (K= 2–18; FCT= 0.3710–0.8719). The number of groups with the highest FCT that included no single population group was four (K= 4). In the AMOVA in which the data were defined as four genetic groups, most of the variation (62.66%) was explained by differences among groups, 29.19% by differences within populations, and 8.15% by differences among populations within groups (Table 3). We found statistically significant levels of genetic differentiation at all hierarchical levels in the analyses. Overall, this group configuration was consistent with the population clusters depicted in the ML tree (Fig. 1).

Table 3.  Results of the analysis of molecular variance (AMOVA) of cpDNA sequence data of Platycarya strobilacea from four genetic groups
Source of variation df Sum of squaresVariance componentsPercentage of variationFixation indices
  1. Significance was tested by 1000 random permutations. ***, P < 0.0001.

Among groups3351.5852.4617***62.66 F CT= 0.626 63
Among populations within groups23171.0131.1465***29.19 F SC= 0.781 67
Within populations153 48.9960.3202*** 8.15 F ST= 0.918 48
Figure 1.

The maximum likelihood tree for Platycarya strobilacea based on chloroplast DNA (cpDNA) atpB-rbcL and psbA-trnH regions variations. The Juglans regia (haplotype 36) was outgroup. Each tip is labeled with the Haplotype, followed by population name in parentheses. Branches are drawn in proportion to the number of substitution per site and measured with the scale. Bootstrap probabilities (BP > 50%) and branch lengths are indicated above the branches. Letters (A, B, C, D) at internal nodes represent clades discussed in the text. Vinculums indicate lineage distribution areas.

2.2 Relationships among haplotypes and divergence time

The aligned length of the two haplotype regions was 1112 bp, including 21 gaps and 36 nucleotide substitutions within P. strobilacea and the outgroup. Of the 57 variable characters, only 30 were parsimony informative characters. Phylogenetic analyses produced 1628 MP trees of 79 steps (CI= 0.5278, RI= 0.6383, RC= 0.3369). In the strict consensus tree, two major sister clades of P. strobilacea with a low bootstrap value were recognized (not shown). By comparison, the ML tree of P. strobilacea was separated into four major clades with moderate to weak supporting values, which is generally consistent with the MP tree but with better resolution (Fig. 1). On the ML tree, fifteen haplotypes in clade A formed a well-supported Southwestern clade, which contained the largest number of haplotypes, and fell near the root of the tree. Within this clade, the relationships among the haplotypes were poorly resolved. Middle-Southern clade B and Eastern clade C, each including six haplotypes, were weakly supported. The monophyletic nature of middle-Western clade D was well supported, possessing eight haplotypes with well resolved relationships. However, clades B, C, and D formed an unresolved polytomy separate from clade A.

The statistical parsimony haplotype network of all samples revealed a single network (Fig. 2). The TCS network was more inclusive and reflected the major clades recovered in the ML tree with a higher degree of precision. Four genetic lineages were indentified by both SAMOVA and phylogenetic analyses, which were marked in the TCS network. Haplotypes 2 and 9 were inferred as the ancestral haplotypes determined by outgroup weight, which was based on both haplotype positions in the network and the phylogeny tree. Many rare haplotypes and a number of missing haplotypes were observed in P. strobilacea. The most recent common ancestors of all major lineages fall within the Eocene and Miocene: 40.06 (HPD 23.41–66.36), 24.49 (HPD 9.08–44.71), 33.30 (HPD 14.21–56.07) and 35.22 (HPD 16.89–57.45) Ma for clades A, B, C and D, respectively. The divergence time between the lineages located in the Southwestern and in the other three areas was estimated to be the late Cretaceous, about 79.78 (HPD 46.32–113.94) Ma before present.

Figure 2.

Minimum spanning network of 35 cpDNA haplotypes in Platycarya strobilacea. The network was rooted at the Juglans regia. Circle sizes are proportional to the number of samples per haplotypes. Solid dots indicate the number of mutational steps. Vinculums indicate haplotype lineages A, B, C and D, which are consistent with that in Fig. 1.

2.3 Geographical distribution of haplotypes

There was a marked cpDNA phylogeographic structure across the 27 P. strobilacea populations, with NST values significantly greater than their corresponding GST values (0.862 vs 0.720, P < 0.01). The 35 haplotypes displayed a rather clear geographical pattern. Moreover, the SAMOVA demonstrates four diverging groups of related haplotypes that correspond fairly well with geography (Fig. 3). Group A occurred in Southwestern China (populations KM, BB, XY, BS, LP, CQ, and ZY), group B in middle-Southern China (populations GX, GL, TD), group C in Eastern China (populations SD, TM, LS, HS, QY, FJ, PS, GD), and group D in middle-Western China (populations NX, HB, JX, AH, HN, WD, SX, GS). The Southwestern group (Group A) contains the most abundant private and rare haplotypes, with population XY (Guizhou) having the highest haplotype diversity in Yungui Plateau (YGP). Populations BB (Guizhou), TD (Hunan), JX (Jiangxi) and GS (Gansu) were also found to have relatively abundant haplotypes. It is surprising that the JX population possessed haplotypes from three different groups. One unique haplotype occurs in the disjunct populations from Jeju Island and Wudang Mountain, and all individuals from the two sites share this haplotype.

Figure 3.

Geographic distribution of the chloroplast DNA haplotypes in Platycarya strobilacea. This map is modified from the website by author. Top left inset: the present distribution of P. strobilacea. Identities of haplotypes lineages in ML tree are represented by different colors. This map is modified from the website AH, Yaoluoping, Anhui, China; BB, Baibao, Wenshan, Yunnan, China; BS, Tianlin, Baise, Guangxi, China; CQ, Jingfe Mt., Chongqin, China; FJ, Longtan, Fujian, China; GD, Zhaoqin, Guangdong, China; GL, Xishan Mt., Guilin, Guangxi, China; GS, Wenxian, Gansu, China; GX, Diecai Mt., Guilin, Guangxi, China; HB, Tuchen, Yichang, Hubei, China; HN, Songshan Mt., Henan, China; HS, Huangshan Mt., Anhui, China; JX, Lushan Mt., Jiangxi, China; KM, Xishan Mt., Kunming, Yunnan, China; KO, Jeju Island, Korea; LP, Liupanshui, Guizhou, China; LS, Baiyun Mt., Lishui, Zhejiang, China; NX, Ningxiang, Hunan, China; PS, Pingshi, Lechang, Guangdong, China; QY, Qingyuan, Lishui, Zhejiang, China; SD, Laoshan Mt., Shandong, China; SX, Zhengpin, Shanxi, China; TD, Lingkou, Tongdao, Hunan, China; TM, Tianmu Mt., Zhejiang, China; WD, Wudan Mt., Hubei, China; XY, Xinyi, Guizhou, China; ZY, Zhunyi, Guizhou, China.

2.4 Demographic history

We observed a mismatch distribution for the cpDNA haplotypes, and calculated each of the four genetic groups separately (Fig. 4). Although groups A and C both showed a relatively smooth, unimodal distribution, only group C (Eastern China) closely matched the expected distribution under an exponential growth rate model, indicating a rapid demographic expansion. The distributions of groups B and D showed some multimodality, a pattern indicative of more deeply diverging lineages. Further, Tajima's D and Fu's Fs tests were both significantly negative for group C (Tajima's D, D=−1.9000, P < 0.05; Fu's, Fs =−3.045 08, P= 0.01) but not for the other groups, indicating that this species experienced a demographic expansion event under a neutral model in group C. Overall, these data provided evidence for recent demographic expansions of P. strobilacea in Eastern China.

Figure 4.

Mismatch distribution plots for major Platycarya strobilacea haplotypes clades. Observed (Obs) frequency of pairwise differences versus expected (Exp) under exponential population growth model for major cpDNA lineages: (A) Southwestern, (B) middle-Southern, (C) Eastern, (D) middle-Western. Corresponding clades are shown in the corner of each panel.

3 Discussion

3.1 High genetic diversity and long evolutionary history

Platycarya, like the well-known maiden hair tree (Ginkgo) and dawn redwood (Metasequoia), is an example of a genus once widespread in the Northern Hemisphere but now endemic to eastern Asia (Manchester et al., 2009). These taxa have undergone long-term evolutionary events and maintained corresponding genetic singles. Despite our limited sampling in terms of the numbers of populations and individuals per population, Platycarya strobilacea shows higher levels of nucleotide diversity (hT= 0.918, πT= 0.0058) than most other flowering plant taxa (i.e. single species, species complexes, or monotypic genera) in East Asian (Liu et al., 2009; Wang et al., 2009; Bai et al., 2010; Qiu et al., 2011). In this study, we were able to distinguish 35 haplotypes by 39 nucleotide substitutions and 18 indels. Compared to previous studies of broadleaved trees (Tomaru et al., 1997; Fujii et al., 2002; Okaura & Harada, 2002; Aoki et al., 2004a, 2004b, 2006; Kanno et al., 2004; Okaura et al., 2007; Hiraoka & Tomaru, 2009a, 2009b), the number of haplotypes we found in P. strobilacea is one of the largest so far for a single tree species in East Asian. Furthermore, we have observed an extremely high genetic diversity of the genus using ISSR marks (Chen et al., 2012, unpublished data). Such high total gene diversity is generally thought to reflect long evolutionary history, which may have allowed accumulation of genetic variation (Huang et al., 2001).

Despite a high level of cpDNA detected at the species level, haplotype variation within populations of P. strobilacea was low (hs= 0.231). Our results revealed significant population subdivision (GST= 0.720) and a strong phylogeographic structure (NST > GST, P < 0.01). This pattern may be attributed to the strong landscape effect of the mountain ranges and climatic conditions as natural dispersal barriers, which greatly reduced gene flow by seed. High gene diversity also suggests that interpopulation gene flow is limited, hence allowing higher levels of total diversity to be maintained (Varvio et al., 1986). In addition, habitat fragmentation resulting from the climatic vicissitudes of the Tertiary and Quaternary may explain its present distribution and restricted gene flow, leading to their high population and/or regional subdivision (Wang & Ge, 2006).

3.2 Multiple refugia

Together with palaeoclimatic and palaeovegetational evidence, our genetic data support the existence of past fragmentation and multiple refugia for P. strobilacea in East Asian. The present phylogenetic analyses and AMOVA analyses jointly support that the cpDNA haplotypes of P. strobilacea belong to four different lineages, which are restricted to the Southwest (YGP), Eastern, middle-Southern and middle-Western China areas, respectively (Fig. 3). A number of unique haplotypes occupy each of these four regions. It is unlikely that these private haplotypes originated in situ during the present interglacial period, given the low mutation rate of cpDNA (Petit & Vendramin, 2007). Moreover, the extent of genetic differentiation between these areas was high (ΦCT= 0.58). These results suggest past fragmentation of this plant species into separate regions, which is consistent with our inference of ancient divergence for these lineages.

The fragmentation events were much older than the last glacial maximum (LGM), and multiple glacial refugia likely existed throughout their current range during Quaternary climatic oscillations (Aoki et al., 2006). Our data provide information about at least four glacial refugia of P. strobilacea located in Southwestern China (YGP), Qinling Mountain, Lu Mountain and Tongdao in Hunan province. Multiple local refugia during late glacial cycles have also been suggested for the Chinese Yew Taxus wallichiana (Gao et al., 2007), an evergreen shrub or tree that has a range similar to that of P. strobilacea. Some of the Chinese Yew's refugia are identical to those identified in the current study (including around Sichuan Basin; E China). Phylogeographic histories of other plant species from subtropical China with habitats broadly similar to those of P. strobilacea also indicate long-term fragmentation and refugial survival. For instance, two glacial refugia of Ginkgo biloba existed in Southwest China and the far east of subtropical China (Gong et al., 2008). Similarly, Eurycorymbus cavaleriei had three major refugia in the southeast of subtropical China (Wang et al., 2009). For another temperate-deciduous walnut tree, Juglans manshurica, two disjunct refugia have been identified within the species’ extant range in northern China (Bai et al., 2010). Noticeably, these findings of multiple refugia are consistent with palaeo-data indicating that subtropical mainland Asia was not covered by major ice sheets during the last glacial period and that the region sustained temperate deciduous forests during the LGM (Ono, 1984; Harrison et al., 2001). With the exception of population JX, none of our sampled P. strobilacea populations yielded strong genetic evidence for an admixture of populations stemming from separate refugia. Our results also do not suggest cleat-cut patterns of postglacial re-colonization from the south as predicted from the fossil pollen data (Yu et al., 2000; Harrison et al., 2001).

The earliest reliable fossils of Platycarya are known from the Early Eocene of England (Manchester, 1987) and North Dakota, USA (Wing & Hickey, 1984). By integrating fossil, morphological and molecular data, Manos et al. (2007) estimated the age of the stem lineage of Platycarya to be 47–65 Ma. We obtained much older estimates, with the most recent common ancestors of haplotypes of extant Platycarya in China dating back to the late Cretaceous (79.78, HPD 46.32–113.94 Ma), and with the divergence of the two major lineages occurring before the Miocene. Although the use of consistent mutation rates may be problematic (Nei, 1992), these estimates offer a relative time frame for the evolution of Platycarya.

3.3 Localized demographic expansion

Due to the lack of samples from Vietnam, analyses of the Northern Indo-China areas are absent in our study. Generally, with their abundant mountains and river gorges, these areas have long been recognized as locations of independent refugia and locally endemic species (e.g. Myricaria laxiflora in the TGMR (Liu et al., 2009)), which is particularly consistent with the exceptional concentration of species diversity in Southwest China, where almost one half of plant species in China are distributed (Wu et al., 1980; Wu & Wu, 1996). We show evidence for the recent population expansion of P. strobilacea in East China, the timing of which remains unknown. The TMRCA of all Eastern lineages was dated to 33.3 Ma before present (HPD, 14.21–56.07), far earlier than the end of the LGM, but we could not accept or reject the postglacial expansion hypothesis.


We thank Byung-Yun SUN (Korea), Zhong-Xin ZHANG, En-Xiang LI, Wu-Sheng JIANG, Hua XIN, Yi REN, Zhi-Xue LIU, Pan LI, Shan LI who assisted with the sample collection of the Paltycarya. We are grateful for technical assistance from Xin JIN. We especially thank Ying-Xiong QIU and two anonymous reviewers for their valuable comments on the early draft of the manuscript, and Feng-Shen HU for helping with English revision. This study was supported by the Natural Science Foundation of Shanghai, China, and the Lishui Administration of Science and Technology Support Program, Zhejiang Province to Shi-Chao Chen (Grant Nos. 10ZR1431300, 20090310), and also by the National Natural Science Foundation of China to Cheng-Xin FU (Grant No. 30830011).