Molecular Phylogenetics and Evolution 85 (2015) 238–246
Contents lists available at ScienceDirect
Molecular Phylogenetics and Evolution
journal homepage: www.elsevier.com/locate/ympev
Repeated range expansions and inter-/postglacial recolonization routes of Sargentodoxa cuneata (Oliv.) Rehd. et Wils. (Lardizabalaceae) in subtropical China revealed by chloroplast phylogeography
Shuang Tian a,b,c,1, Shu-Qing Lei b,1, Wan Hu a,b,1, Ling-Li Deng b,BoLib, Qing-Lin Meng b, ⇑ ⇑ Douglas E. Soltis d,e, Pamela S. Soltis e, Deng-Mei Fan b, , Zhi-Yong Zhang b, a College of Forestry, Jiangxi Agricultural University, 330045 Nanchang, Jiangxi, China b Laboratory of Subtropical Biodiversity, Jiangxi Agricultural University, 330045 Nanchang, Jiangxi, China c Jiangdezhen College, 333000 Jingdezhen, Jiangxi, China d Department of Biology, University of Florida, Gainesville, FL 17 32611, USA e Florida Museum of Natural History, University of Florida, Gainesville, FL 17 32611, USA article info abstract
Article history: Most plant phylogeographic studies in subtropical China have stressed the importance of multiple refugia Received 18 December 2014 and limited admixture among refugia. Little attention has been paid to range expansion and recoloniza- Revised 15 February 2015 tion routes in this region. In this study, we implemented a phylogeographic survey on Sargentodoxa Accepted 18 February 2015 cuneata, a widespread woody deciduous climber in subtropical China to determine if it conforms to Available online 27 February 2015 the expansion–contraction (EC) model during the Pleistocene. Sequence variation of two chloroplast intergenic spacers (IGSs) in 369 individuals from 54 populations of S. cuneata was examined. Twenty- Keywords: six chloroplast haplotypes were recovered. One of these (H5) occurred across the range of S. cuneata Range expansion and was absent from only 13 populations. Sixteen of the 26 haplotypes were connected to H5 by one Recolonization Chloroplast mutation and displayed a star-like pattern in the haplotype network. All chloroplast haplotypes clustered Phylogeography into two lineages (A and B) in a Bayesian tree, and most haplotypes (18 out of 26) originated during the mid-Pleistocene (0.63–1.07 Ma). Demographic analyses detected a recent range expansion that occurred at 95.98 ka (CI: 61.7–112.53 ka) for Lineage A. The genetic signature of an ancient range expansion after the Middle Pleistocene Transition (MPT) was also evident. Three recolonization routes were identified in subtropical China. The results suggest that temperate plants in subtropical China may conform to the EC model to some extent. However, the genetic signature from multiple historical processes may complicate the phylogeographic patterns of organisms in the region due to the mild Pleistocene climate. This study provides a new perspective for understanding the evolutionary history of temperate plants in subtropical China. Ó 2015 Elsevier Inc. All rights reserved.
1. Introduction areas of both continents uninhabitable (reviewed by Hewitt, 1996, 2000; Shafer et al., 2010; Soltis et al., 2006). During glacial periods, Most extant plant species from the Northern Hemisphere have temperate species of these regions were driven southward but persisted through glacial–interglacial cycles over the past 2.5 then expanded their distributions northward during warmer million years. Range shifts are the most conspicuous response of interglacial/postglacial periods. However, the range shift of plant taxa to Pleistocene climatic fluctuations (Davis and Shaw, temperate plants in subtropical China is much more complex, only 2001), especially for temperate species in Europe and North several cases found regional expansions (e.g., Lei et al., 2012; Qi America, where the development of massive ice sheets made large et al., 2012; Sun et al., 2014) but few conform to a range-wide expansion–contraction (EC) model that is typical for many species from Europe and North America (Harrison et al., 2001; Liu et al., ⇑ Corresponding authors at: No. 1101, Zhimin Road, Nanchang, Jiangxi 330045, 2012; Ni et al., 2010; Qiu et al., 2011). PR China. Fax: +86 791 83813047. The hilly mid-elevation area of eastern China between the E-mail addresses: [email protected] (D.-M. Fan), [email protected] Qinling Mts. Huai River line (at C. 34°N) and the tropical South (Z.-Y. Zhang). 1 6 These authors contributed equally. ( 22°N), and bordered by the Qinghai-Tibetan Plateau (105°E) in http://dx.doi.org/10.1016/j.ympev.2015.02.016 1055-7903/Ó 2015 Elsevier Inc. All rights reserved. S. Tian et al. / Molecular Phylogenetics and Evolution 85 (2015) 238–246 239 the west and the coastline in the east, is generally referred to as extant member, S. cuneata (Oliv.) Rehd. et Wils., is a deciduous ‘subtropical China’ (Zhao, 1986). This area now harbors warm- woody climber, mostly confined to subtropical China with occa- temperate evergreen forests (i.e., subtropical evergreen broad- sional occurrences in Laos and northern Vietnam (Chen and leaved forests) which are interspersed with warm-temperate Tatemi, 2001). Plants of S. cuneata frequently grow in different deciduous forests (Harrison et al., 2001; Wu, 1980). Although sub- types of forests with sufficient sun and in thickets at forest margins tropical China was probably not glaciated during the Pleistocene from 130 m to 2400 m above sea level (a.s.l., Ying et al., 1993). This (Shi, 1998), the region underwent profound climate changes species bears fleshy, dark blue berries (Chen and Tatemi, 2001), throughout the Pleistocene. At the Last Glacial Maximum (LGM), which may facilitate its dispersal by birds. Given the wide dis- for example, the climate of this region was cooler by c. 4–6 °C tribution of the species in subtropical China, its occurrence in and drier by c. 400–600 mm/yr (Qiu et al., 2011). Paleoecological diverse habitats, and efficient seed-dispersal, S. cuneata may repre- reconstructions based on fossil pollen showed that marked climate sent a good system to test if there are some temperate plants that change caused warm-temperate evergreen forests to retreat south- conform to the EC model in subtropical China. If so, this species ward c. 1000 km relative to today (Harrison et al., 2001; Ni et al., would provide a good opportunity to identify the recolonization 2010; Yu et al., 2000). However, the temperate deciduous forests routes in subtropical China following the Pleistocene glaciations. retreated to the continental mountains and were separated by We investigated the phylogeographic structure of S. cuneata in more cold-tolerant biomes in the lowlands, such as boreal, cool- subtropical China using chloroplast DNA sequence data. The main temperate coniferous, and mixed (boreal conifers with the most objectives were: (i) to examine the phylogeographic structure of S. cold-tolerant temperate taxa) forests. After the Holocene, warm- cuneata; (ii) to investigate signals of range expansion(s) and to temperate evergreen forests re-colonized the mid-latitude lowland attempt to peel through different layers of historical processes; areas up to 30/33°N. The expansion of evergreen forests displaced and (iii) to identify potential recolonization routes in subtropical temperate deciduous forests over the same area, resulting in cur- China for temperate plant species. rently isolated patches of temperate deciduous forest at both higher elevations and more northern latitudes (Harrison et al., 2. Materials and methods 2001; Ni et al., 2010; Yu et al., 2000). The history of temperate deciduous forests in subtropical China 2.1. Population sampling and experimental procedures indicates that temperate plant species in this region may have been isolated during both glacial and inter-/postglacial stages, We made extensive field surveys throughout subtropical China giving rise to an expected phylogeographic pattern of high diver- from 2010 to 2014 and collected leaf samples from 584 individuals sity among populations due to genetic drift and limited genetic in 54 populations (Fig. 1 and Supplementary Table 1). Because the admixture between presently disjunct populations. A few recent species is a very long vine (up to five meters or more), sampled studies (e.g., Gong et al., 2008; Lei et al., 2012; Qiu et al., 2009; individuals were spaced by ca. 20–100 m to avoid repeatedly col- Zhang et al., 2013) have suggested that the phylogeographic struc- lecting the same individual. All samples were desiccated in silica tures of temperate plants in subtropical China are indeed different gel and stored at À20 °C until being processed. Genomic DNA from those of their counterparts in North America and Europe. was extracted using a modified CTAB procedure. These findings match the predictions of high population differ- After a screening of a dozen of intergenic spacers (IGSs) of the entiation and limited genetic admixture due to long-term pop- chloroplast genome, two IGSs, atpI–atpH and trnS–trnG, had high ulation isolation (Harrison et al., 2001; Ni et al., 2010; Yu et al., sequence variation and were selected as molecular markers for this 2000). However, both paleoecological reconstructions and previous study. PCR amplification reactions using previously reported phylogeographic studies have limitations, which may lead to an primers (Shaw et al., 2007) were carried out in a volume of 20 ll incomplete understanding of the evolutionary history of temperate containing 10 ll2Â Taq PCR MasterMix (Tiangen, Shanghai, plants in subtropical China. First, paleoecological reconstructions China), 1 ll each forward and reverse primer (0.2 lM), 1 ll tem- of entire forest biomes cannot provide a detailed picture of past plate DNA (ca. 50–100 ng) and 7 ll ddH O. Amplification was car- fragmentation, admixture, and/or expansion of populations of par- 2 ried out in a Bioer XP cycler (Bioer, Hangzhou, China) programmed ticular species (Qian and Ricklefs, 2001). Second, fossil pollen for an initial 240 s at 94 °C, followed by 30 cycles of 60 s at 94 °C, records used for such biome reconstructions are largely restricted 60 s at 54 °C(atpI–atpH)or57°C(trnS–trnG), 60 s at 72 °C, and a to the most recent (LGM/Holocene) time periods and are limited final 600 s at 72 °C. Sequencing reactions were conducted with in both taxonomic resolution and spatial coverage, especially in the corresponding forward and reverse primers commercially by China (e.g., Harrison et al., 2001; Yu et al., 2000). Third, because Sangon Biotech Co., Ltd. (Shanghai, China). many plants have wide ecological requirements and are not con- fined to a specific kind of vegetation (referred to as ‘generalists’), they may have had greater opportunities for population expansion, 2.2. Data analysis possibly conforming to the EC model. However, temperate general- ist plants have rarely been considered in previous phylogeographic 2.2.1. Genetic diversity and genetic structure studies. Fourth, due to the relatively mild Pleistocene climate in Because the chloroplast genome is regarded as a single locus, subtropical China, different glacial–interglacial cycles could have the two sequenced fragments were concatenated to define resulted in multiple layers of genetic signature This would yield cpDNA haplotypes. Total genetic diversity (hT) and within- a complex phylogeographic structure, which in turn would compli- population diversity (hS) were calculated with HAPLONST (Pons cate phylogeographic inference. However, few studies in this and Petit, 1996). Phylogeographic signal in the haplotype dis- region have attempted to disentangle the contributions of these tribution was evaluated by comparing GST with NST using the soft- multiple genetic layers. ware HAPLONST (excluding populations with sample sizes less
Sargentodoxa is a monotypic genus with a relictual distribution than 3). A higher NST than GST usually indicates the presence of (Chen and Tatemi, 2001). Because of a number of autapomorphies, phylogeographic structure (Pons and Petit, 1996), with closely it has been treated as a monotypic family, Sargentodoxaceae (Stapf, related haplotypes being found in the same area more often than 1926). However, Wang et al. (2009a) suggested that Sargentodoxa less closely related haplotypes. A haplotype network was inferred is best retained as a member of Lardizabalaceae as it shares many under the criterion of statistical parsimony using Network 4.6 synapomorphies with the remaining Lardizabalaceae. The only (Bandelt et al., 1999) using Akebia trifoliata (Thunb.) Koidz 240 S. Tian et al. / Molecular Phylogenetics and Evolution 85 (2015) 238–246
Fig. 1. Distribution of all chloroplast haplotypes of Sargentodoxa cuneata. (A) Distribution of the dominant haplotype, H5 (dot), three major recolonization routes of S. cuneata and the network of 26 chloroplast haplotypes and outgroup, Akebia trifoliata (bottom-right inset). (B). Purple arrows: Eastern Route; Red arrows: Middle Route; Blue arrows: Western Route.
(Lardizabalaceae) as outgroup based on the phylogeny of Eupteleaceae, and two in Berberidaceae) using matK gene Lardizabalaceae (Wang et al., 2002). Haplotype distribution maps sequences that are available from GenBank (Supplementary were constructed using ArcGIS 9.3. Table 2). S. cuneata was represented by four accessions. Bayesian searches for tree topologies and node ages of Ranunculales were 2.2.2. Divergence time estimation of lineages and substitution rates of conducted in BEAST (Version 7.1; Drummond and Rambaut, two IGSs 2007) using the HKY substitution model, selected by MODELTEST To estimate the divergence time of lineages and specify a sub- (Version 3.6; Posada and Crandall, 1998). A Yule process was speci- stitution rate for the two IGSs of S. cuneata, we used a secondary fied as the tree prior. Four fossil points were used to assign mini- calibration. First, we estimated the crown group age of S. cuneata mum age constraints on four internal stem nodes: node A, B, C, by reconstructing the phylogeny of 15 members of Ranunculales and D (Fig. 2). The justification of these fossils are given in (five genera in Lardizabalaceae, five in Menispermaceae, one in Supplementary Table 3. Assuming a normal distribution, these S. Tian et al. / Molecular Phylogenetics and Evolution 85 (2015) 238–246 241
Fig. 2. BEAST-derived chronogram of Ranunculales based on matK sequences. Four fossil dates: (A) 91 Ma (Knobloch and Mai, 1986); (B) 89.3 Ma (Knobloch and Mai, 1986); (C) 55.2 Ma (Jacques and De Franceschi, 2005); (D) 33.9 Ma (Jacques and Guo, 2007). Gray bars on each node indicate 95% highest posterior densities (HPDs) of time estimates (Ma). The crown group age of S. cuneata is 7.29 (2.86–13.26) Ma (Node E). All clades received 100% posterior probability support. calibration points were modeled with means of 91, 89.3, 55.2 and not be rejected at the 5% level, an estimate of the parameter of 33.9 Ma (SD = 1.0), respectively. the spatial or demographic expansion was obtained. In addition, Second, a BEAST analysis was conducted using all cpDNA-IGS we used tests of selective neutrality (Fu’s (1997) Fs; Tajima’s haplotype sequences of S. cuneata to determine intraspecific node (1989) D) to infer potential population growth and expansion. ages. We used the same settings as in the first step, except for a For total populations and Lineage A that had signals of expansion HKY substitution model and a constant-size coalescent tree prior. (see Section 3), the expansion parameter (s), and its 95% confidence To calibrate the root node, we used the crown group age (median interval (CI), were converted to absolute estimates of time (T,in value) of S. cuneata estimated from a broader Ranunculales phy- number of generation) since expansion began using T = s/2u logeny, that is 7.29 (2.86–13.26) Ma. Akebia trifoliata was again (Rogers, 1995; Rogers and Harpending, 1992), where u is the neu- selected as the outgroup of the S. cuneata haplotypes. tral mutation rate for the entire sequence per generation. The value For each BEAST analysis, MCMC runs were performed, each of for u was calculated as u = lkg, where l is the substitution rate in 107 generations, with sampling every 1000 generations, with a substitutions per site per yr (s/s/y) that was obtained from the burn-in of the initial 20% of the cycles. MCMC samples were clock-calibrated BEAST tree of all haplotypes of S. cuneata, k is the inspected in TRACER (version 1.5) to confirm sampling adequacy aligned sequence length of the two IGSs (1491 bp, excluding a and convergence of the chains to a stationary distribution. 9 bp indel, see Results), and g is the generation time in yr (i.e. age Resulting chronograms were visualized in FIGTREE (version 1.3.1; of first reproduction). There is no accurate record for the first repro- http://tree.bio.ed.ac.uk/software/figtree/). duction age of S. cuneata, but some woody climbers of the same family, such as Akebia trifoliata (Thunb.) Koidz., have been culti- vated. Plants of A. trifoliata first flower in the second year, but their 2.2.3. Demographic analysis flowering increases significantly in the third year (Xiong et al., The observed patterns of pairwise difference between haplo- 2007). Therefore, we used 3 yr as the generation time for S. cuneata. types within populations (mismatch distribution) were compared to those obtained under a sudden demographic expansion or a spa- tial expansion model using ARLEQUIN for total populations and 3. Results major lineages (Lineage A and Lineage B, see Results) identified in haplotype phylogeny reconstruction, respectively. Goodness of 3.1. Sequence variation, genetic diversity, and genetic structure fit based on the sum of the squared differences (SSD) was tested using parametric bootstrapping with 1000 pseudoreplicates We obtained sequences of the atpI–atpH and trnS–trnG inter- (Excoffier et al., 2006) for each model. Each time a model could genic spacers (IGSs) from 369 individuals; some of the plants 242 S. Tian et al. / Molecular Phylogenetics and Evolution 85 (2015) 238–246 originally collected did not yield results because S. cuneata con- cuneata calculated in the matK analysis. Interestingly, most haplo- tains abundant polysaccharides that prevented extraction of types (18 out of 26) originated during the mid-Pleistocene (0.63– high-quality genomic DNA in some samples and posed many 1.07 Ma, grey-shaded column in Fig. 3), implying this period is difficulties for PCR reactions. The aligned sequences of atpI–atpH extremely important for the differentiation of S. cuneata. Based on and trnS–trnG from 54 populations of S. cuneana were 776 bp and the cpDNA-IGS chronogram, the BEAST analysis provided an aver- 724 bp in length, respectively. Six and 17 substitutions were found age substitution rate of 0.75 Â 10À9 s/s/y for the two cpDNA-IGSs. in atpI–atpH and trnS–trnG, respectively. One indel (9 bp) in trnS– trnG was discarded in subsequent analyses, resulting in a reduced 3.4. Demographic history length of the combined sequences (1491 bp) for further analyses (Supplementary Table 4). New sequences of S. cuneata along with The unimodal distribution pattern of pairwise differences those of outgroup have been deposited in GenBank under acces- (Supplementary Fig. 1), non-significant SSD, and significant sion numbers KP658762-KP658780, KP658803-KP658810, and Tajima’d D and Fs (Table 1) showed that both the total populations KP771881-KP771882 (Akebia trifoliata). and Lineage A underwent recent range expansion events. However, Analyses of the combined sequences of the two fragments per- s values of the total populations differed significantly between the mitted the identification of 26 haplotypes among all individuals spatial and demographic expansion models and had large confi- examined. Total genetic diversity h (0.665) across all populations T dence intervals. Therefore, we did not consider the s values of total was much higher than average within-population diversity h S populations as a reliable inference of expansion time. In contrast, s (0.246), resulting in high population differentiation (G = 0.630, ST values of Lineage A under both models were very similar. Under N = 0.603). A permutation test indicated that N was not signifi- ST ST the spatial expansion model, the s value yielded a time estimate cantly higher than G (P = 0.147). ST of range expansion of 95.98 ka (CI: 61.7–112.53 ka), which corre- sponds well to the last interglacial in China (75–100 ka, Marine 3.2. Haplotype network and haplotype distribution Isotope Stage, MIS 5, Cui et al., 2011).
The network (Fig. 1B) of 26 haplotypes displayed a star-like pattern, with 16 haplotypes directly connecting to H5 by one 4. Discussion mutation. A second minor center is H19, connecting to 4 haplo- types (H20–H23), each by one mutation. Only a few haplotypes 4.1. Repeated range expansions during inter-/postglacial periods (H2, H3, H24, H25, H26) were relatively removed from H5 or H19 (by two to four mutations). Obviously, H5 is an ancestral hap- Obvious signals of extensive range expansion were recovered lotype because it was closely related to the outgroup and occupied for S. cuneata in subtropical China by surveying chloroplast DNA an interior position in the network. sequence variation. These results, coupled with a few recent Of 54 populations of S. cuneata, 31 were fixed for a single reports (Li et al., 2012b; Qi et al., 2012; Sun et al., 2014), stress haplotype, while the remaining 23 were polymorphic the importance of range expansion in shaping the phylogeographic (Supplementary Table 1 and Fig. 1A). The polymorphic populations structure of plant species in subtropical China, shedding new were concentrated mostly in southern, central, and eastern areas, insights into the complex evolutionary history of temperate plants particularly in the southeastern Yungui Plateau and Nanling in this region. Mountains (18, 21, 24, 25, 34 and 35), Jiuling Mountains (22 and In fact, range expansion for temperate plants in subtropical 29), Luoxiao Mountains (20, 22, 30, 31 and 33), and Wuyi China has been heavily underestimated in previous studies. Mountains (39, 41, 43 and 48). Other polymorphic populations Although fossil pollen data suggest that temperate deciduous for- were interspersed across the range of S. cuneata; these include ests persisted in continental mountains between 22°Nto34°N one western population (4), one southwestern population (9), (Harrison et al., 2001; Ni et al., 2010; Yu et al., 2000), this veg- two northernmost populations (44 and 45), and one easternmost etation may also have retreated to the south as did other types population (52). The most common haplotype, H5, occurred across of vegetation when the climate became cooler and drier. For exam- the range of S. cuneata, and was absent in only 13 populations ple, pollen sampled from the Yangtze River Delta from the late (Fig. 1 and Supplementary Table 1). Ten haplotypes were regionally Pleistocene to early Holocene (Younger Dryas) is dominated by abundant or found in multiple populations. For example, H7 the xerophytic plants Artemisia and common conifers, whereas occurred in five northeastern populations (40, 41, 44, 45 and 50), from 10,300 to 9000 yr BP, mixed broadleaved evergreen-decidu- H12 in three eastern populations (52, 53 and 54), H14 in two pop- ous forests developed on the grasslands and surrounding hills ulations of Yungui Plateau (4 and 10), H19 in three southwestern and uplands (Yi et al., 2003). Furthermore, there is a large non- populations (9, 18 and 26) and one population of Jiangxi forest area in central China that is the result of lower temperature Province (39), and H21 in three southeastern populations (32, 35 and drier climate during the LGM (Harrison et al., 2001). This find- and 36). Fifteen other haplotypes (H3, H4, H6, H8, H9, H10, H11, ing indicates that temperate deciduous forests now occupying the H13, H15, H16, H17, H22, H23, H24 and H26) were private to a sin- non-forest area could be a consequence of range expansion during gle population. the warmer and moister Holocene. Therefore, the conclusion of some researchers that there was very little inter-/postglacial range 3.3. Divergence time and substitution rate of two IGSs haplotypes expansion may reflect the fact that the early phylogeographic stud- ies in subtropical China were biased toward endangered species The matK chronogram of Ranunculales derived from the BEAST with narrow distributions (e.g., Gong et al., 2008; Li et al., 2012a; analysis recovered Sargentodoxa as sister to all other members of Wang and Ge, 2006; Wang et al., 2009b; Zhou et al., 2010) or plants Lardizabalaceae with high support (posterior probability, that only inhabited temperate forests (Chen et al., 2012a,b; Lei PP = 100%) (Fig. 2). The crown group age of Sargentodoxa was et al., 2012; Qiu et al., 2009; Zhang et al., 2013). And this conclu- 7.29 Ma (95% HPD: 2.86–13.26 Ma) based on the matK chronogram sion might also result from the overlook of the multiple genetic (Fig. 2). BEAST analysis showed that all IGS haplotypes clustered signatures of different processes (Zhan and Fu, 2011) in previous into two lineages (A and B), which diverged from each other during studies (e.g., Lei et al., 2012; Zhang et al., 2013). the Pliocene (3.16 Ma, 95% HPD: 1.42–7.2 Ma, Fig. 3), using the Recently, range expansion has been detected in several other crown group age as root prior for the cpDNA-IGS chronogram of S. phylogeographic studies. For example, the warm-temperate S. Tian et al. / Molecular Phylogenetics and Evolution 85 (2015) 238–246 243
Fig. 3. The phylogenetic relationships of chloroplast haplotypes of S. cuneata. The age estimate (mean [95% highest posterior density, HPD]) for each node is showed besides the nodes. The lineages with more than 95% posterior probability are indicated by thick branches. An average substitution rate (0.75 Â 10À9 s/s/y) for atpI–atpH and trnS–trnG sequences was provided in this analysis. All chloroplast haplotypes clustered into two Lineages A and Lineages B. The grey-shaded column shows divergence events that happened during the middle Pleistocene (1.07–0.63 Ma).
Table 1 Probabilities for a spatial or a demographic expansion, neutral tests computed on the combined chloroplast data for Total, Lineage A and Lineage B of S. cuneata.
Lineage Spatial expansion Demographic expansion Fu’s Fs Tajima’s SSD p s SSD p s D Total 0.0009 0.5010 0.4600 (0.3540–2.5930) 0.0012 0.7560 0.8750 (0–2.2500) À25.4510⁄ À1.9040⁄ Lineage A 0.0006 0.1220 0.6440 (0.4140–0.7550) 0.0009 0.3330 0.6170 (0.4880–1.0740) À17.6390⁄ À1.9050⁄ Lineage B 0.0358 0.1190 3.0540 (0.7230–5.0240) 0.0370 0.0870 3.4120 (0.7950–5.9120) À0.8530 0.5200
⁄ P < 0.05
Cercidiphyllum japonicum underwent postglacial northward expan- above present sea level (Dutton and Lambeck, 2012). Therefore, the sion in the Sichuan Basin and/or the middle Yangtze (Qi et al., plants could have expanded their ranges in subtropical China dur- 2012). Sun et al. (2014) recently reported the chloroplast-based ing the LGI. However, the possibility of postglacial (Holocene) phylogeography of Tetracentron sinense, in which they also range expansion cannot be ruled out for S. cuneata, particularly detected two independent events of range expansion. Therefore, in the west where H5 dominates most populations. Failure of accumulating evidence suggests that range expansion may have detecting Holocene events is a major limitation for chloroplast played a much more important role in molding the phylogeo- phylogeography, because nucleotide polymorphisms in the gen- graphic pattern of temperate plants in subtropical China than ome mostly predate the LGM (Gavin et al., 2014, and also see cases previously thought, particularly for generalist plants which may in Li et al., 2012b; Qi et al., 2012). Further studies using high- have more opportunities for dispersal due to their wide ecological throughput sequencing to resolve postglacial phylogeography requirements. (e.g., Emerson et al., 2010) are needed to infer the complete history Through mismatch analysis, this study found that S. cuneata of temperate plants in subtropical China. could have experienced a recent range expansion during the last In addition to the LGI range expansion, different lines of evi- glacial interglacial (LGI, 75–100 ka, MIS 5, Cui et al., 2011), consis- dence showed that S. cuneata could have experienced an ancient tent with the findings of a recent phylogeographic study of expansion event after the mid-Pleistocene climate transition Pteroceltis tatarinowii, a widespread temperate canopy tree in (MPT, 0.8–1.2 Ma, Tzedakis et al., 2009), a period marked by an mainland China, which also revealed a range expansion event dur- increase in the severity of glaciations, the emergence of the ing the LGI (Li et al., 2012b). Paleoclimatological studies suggested 100-kyr glacial cycles, with maximum glaciations and warmer that the LGI had a climate as warm as or warmer than today (Kukla interglacials becoming established at the end of the MPT (Clark et al., 2002), with a global (eustatic) sea level peak of 5.5–9 meters et al., 2006). First, an overall star-like network indicated a range 244 S. Tian et al. / Molecular Phylogenetics and Evolution 85 (2015) 238–246 expansion and this event was mainly associated with an ancestral 4.2. Recolonization routes in subtropical China haplotype (H5). This pattern can only be explained by an ancient rather than a recent expansion because the time since the LGI Although range expansions were detected in several phylogeo- expansion (95.98 ka) found in mismatch analysis is not sufficient graphic studies (e.g., Qi et al., 2012; Sun et al., 2014), recolonization to produce the most abundant derived haplotypes, given that the routes for plant species in subtropical China have rarely been substitution rate of the two IGSs is 0.75 Â 10À9 s/s/y. Second, the explored. In this study, three major recolonization routes can be earliest and largest glaciation in China was reported shortly after identified tentatively by visual inspection of the haplotype dis- the MPT. The ice sheet during Wangkun glaciation (700–500 ka) tribution map (Fig. 1). The Eastern Route is from the Wuyi was 18 times larger than the present ice sheet on the Qinghai- Mountains to Tianmu Mountain and the mountains of eastern Tibetan Plateau (Shi, 1998). This, again, almost coincides with the Zhejiang (Purple arrows in Fig. 1B). Many phylogeographic studies time of origin of most haplotypes (0.63–1.07 Ma, Fig. 3). These in subtropical China suggest that the southeastern coastal moun- facts suggest that severe glacial climate after the MPT could have tains (the mountains in Zhejiang and Fujian Provinces along the pushed S. cuneata further south (e.g., Nanling mountains and coastline, including Wuyi Mountains and Tianmu Mountian) are eastern Yungui Pleateau) and led to a highly fragmented dis- a major glacial refugium for different organisms (Ding et al., tribution, which provided opportunities for lineage differentiation 2011; Gong et al., 2008; Wang et al., 2009b) because these moun- and subsequent range expansion during the large interglacial. tains can intercept moisture and heat transported from the ocean, Recently, a few studies began to recognize the importance of providing climatic and ecological stability. However, extending in a the MPT in the evolutionary history of Chinese organisms. For northeast-southwest direction, these mountains (particularly the example, cold climate during the Lonian stage (0.6 Ma) caused Wuyi Mountains) may also have acted as a corridor for southward the differentiation of Chinese populations of Sus scrofa (wild pig) retreat and northward expansion during the Pleistocene. For exam- into Northern and Southern lineages, after which both lineages ple, Chen et al. (2012b) found that a clade of pollinators of Ficus underwent increases in population sizes during the subsequent pumila expanded its range northward along the Wuyi Mountains. warm period (Frantz et al., 2013; Groenen et al., 2012). These find- Our study also found that H5 is widespread in the Wuyi ings are also congruent with recent palynological studies during Mountains and the mountains of Zhejiang Province, suggesting and after the MPT in subtropical China. For example, broad-leaved that S. cuneata may also have used the Wuyi Mountains as a migra- deciduous plants (Carya, Juglans, Platanus, and others) were tion corridor. frequent in north Jiangsu Province during MIS 17–19, then Glacial refugia in the Nanling Mountains and the southern disappeared during MIS 16; however, during MIS 13–15, the Luoxiao Mountains have been revealed by several phylogeographic warm-adapted elements recolonized the area (Niu, 2013). studies (Qiu et al., 2011; Wang et al., 2009b). However, the recolo- More importantly, repeated range expansions help resolve a nization route from the Nanling Mountains, along the Luoxiao longstanding phylogeographic conundrum in subtropical China Mountains to the Jiuling Mountains, and further to Dabie that also exists in S. cuneata. Multiple isolated refugia are common Mountain, has seldom been detected. In this study, H5 is frequent for plant species in subtropical China; however, the same haplo- in this chain of mountains, evidently indicative of a recolonization type or closely related haplotypes are often shared among multiple route (the Middle Route, red arrows in Fig. 1B). Note that along the isolated refugia, which means genetic affinity among refugia (e.g., Middle Route, as well as the Eastern Route, there are many derived Haplotype B in Chou et al., 2011; H14 and H19 in Gao et al., haplotypes (H6, H16, H17 and H18) and polymorphic populations 2007;H4inLei et al., 2012;H8inQiu et al., 2009;H1inShi (20, 29, 30, 39, 41, 43, 44, 45, 48 and 52), suggestive of long-term et al., 2014; Haplotype J in Wang et al., 2009b; Zhang et al., in situ survival after an ancient range expansion. This finding indi- 2013; Haplotype A in Zhou et al., 2010). For example, Lei et al. cates that the mountains along the Eastern and Middle routes (2012) identified four glacial refugia; however, a common and could have acted as both migration corridors and glacial refugia, ancestral haplotype (H4) occurred in all of the four isolated refugia. depending on the severity of the climate during the glaciations. What is the mechanism underlying the contradiction between The Western Route is from the western Nanling Mountains, long-term isolation and shared ancestral haplotype? Long-distance along the eastern Yungui Plateau to the southeast Sichuan Basin dispersal and incomplete lineage sorting are two possible answers. and Three-Gorge Mountains, and from the latter further to the However, the former needs an assumption that rare dispersal Qinling Mountains (Blue arrows in Fig. 1B). In addition, population events happened multiple times, a scenario that may be highly 9 also contains H5, indicative of a branch pointing to the southwest unlikely. The latter seems to be a better explanation, but this (Fig. 1B). In contrast to the Eastern and Middle routes, only a few hypothesis requires a premise that the ancient haplotype (H4) populations contained multiple or derived haplotypes along the should spread across the range prior to the fragmentation, which Western Route, suggesting a more recent expansion event associ- inevitably invokes an ancient range expansion. Multiple refugia ated with H5. In fact, recent range expansion in these areas has and haplotype sharing among refugia also appear in S. cuneata. been revealed by several studies. For example, Sun et al. (2014) Private haplotypes in different parts of the range (e.g., H12 in the recently found that clades 2 and 3 of Tetracentron sinense east, H18 in the northwest, H19 in the southwest) indicate that underwent range expansions in eastern Yungui Plateau and the S. cuneata could have survived through glacial periods in situ in Three-Gorge area, respectively. It is possible that climate changes remote refugia across the range. However, H5 is shared among dif- in the late Pleistocene in this area may have been more severe than ferent glacial refugia, within which different derived haplotypes in southeastern China, perhaps erasing the genetic footprints of were produced. This phenomenon cannot be interpreted by a previous events and displaying a clear signature of a recent range MIG or Holocene expansion because most derived haplotypes of expansion. H5 are dated to the middle Pleistocene. The most likely scenario may be that an ancient expansion event (during MIS 13–15) brought H5 to different parts of the range, and numerous derived 5. Conclusions haplotypes were produced in these isolated locations (Fig. 1A). Subsequently, different haplotypes could have survived in situ Although the analyses of this study were based only on chloro- through the following glacial periods, and several may have been plast markers, the strong range-wide expansion signals found in S. involved in local range expansions during the MIG or Holocene cuneata, coupled with increasing evidence revealed in other stud- (e.g., H7, H12, H21) together with further expansion of H5. ies, stress the importance of the expansion–contraction (EC) model S. Tian et al. / Molecular Phylogenetics and Evolution 85 (2015) 238–246 245 for interpreting the phylogeographic structure of plant species in Gavin, D.G., Fitzpatrick, M.C., Gugger, P.F., et al., 2014. Climate refugia: joint subtropical China. We also found that the genetic legacy of at least inference from fossil records, species distribution models and phylogeography. New Phylot. 204, 37–54. two expansion events was preserved in S. cuneata, implying that Gong, W., Chen, C., Dobes, C., Fu, C.X., Koch, M.A., 2008. Phylogeography of a living different layers of historical processes should be taken into account fossil: Pleistocene glaciations forced Ginkgo biloba L. (Ginkgoaceae) into two in future phylogeographic studies for subtropical Chinese organ- refuge areas in China with limited subsequent postglacial expansion. Mol. Phylogenet. Evol. 48, 1094–1105. isms (Zhan and Fu, 2011). Furthermore, three recolonization routes Groenen, M.A.M., Archibald, A.L., Uenishi, H., et al., 2012. Analyses of pig genomes were tentatively identified for the first time for subtropical Chinese provide insight into porcine demography and evolution. Nature 491, 393–398. plants, which may have significant implications for constructing Harrison, S.P., Yu, G., Takahara, H., Prentice, I.C., 2001. Palaeovegetation (Communications arising): Diversity of temperate plants in east Asia. Nature ecological corridors in subtropical China under global climate 413, 129–130. warming. Hewitt, G.M., 1996. Some genetic consequences of ice ages, and their role in divergence and speciation. Biol. J. Linn. Soc. 58, 247–276. Hewitt, G.M., 2000. The genetic legacy of the quaternary ice ages. Nature 405, 907– Acknowledgments 913. Jacques, F.M.B., De Franceschi, D., 2005. Endocarps of Menispermaceae from Le Quesnoy outcrop (Sparnacian facies, Lower Eocene, Paris Basin). Rev. Palaeobot. The authors are grateful to the people who helped us in the Palyno. 135, 61–70. field. Special thanks go to Yongjun Chen, Shupeng Dong, Yi Yang, Jacques, F.M.B., Guo, S.X., 2007. Palaeoskapha sichuanensis gen. et sp. nov. Yi Luo, Tao Li, Wei Zeng, and Bochao Li for their participation in (Menispermaceae) from the Eocene Relu Formation in western Sichuan, West China. Acta Phytotaxon. Sin. 45, 576–582. sample collection. We are also indebted to Jingping Chen, and Knobloch, E., Mai, D.H., 1986. Monographie der Fruchte und Samen in der Kreide Min Zheng for their laboratory assistance. This study was sup- von Mitteleuropa. Rozpravy Ustrednıhoustavu geolog ickeho 47, 1–279. ported by grants from the National Natural Science Foundation of Kukla, G.J., De Beaulieu, J.L., Bond, G., et al., 2002. Last interglacial climates. Quaternary Res. 58, 2–13. China (Grant Nos. 41061006 and 31160043) and National Science Lei, M., Wang, Q., Wu, Z.J., López-Pujol, J., Li, D.Z., Zhang, Z.Y., 2012. Molecular and Technology Support Program (2012BAC11B02). phylogeography of Fagus engleriana (Fagaceae) in subtropical China: limited admixture among multiple refugia. Tree Genet. Genomes 8, 1203–1212. Li, J., Jin, Z., Tan, T., 2012a. Genetic diversity and differentiation of Sinocalycanthus chinensis populations revealed by chloroplast microsatellite (cpSSRs) markers. Appendix A. Supplementary material Biochem. Syst. Ecol. 41, 48–54. Li, X.H., Shao, J.W., Lu, C., Zhang, X.P., Qiu, Y.X., 2012b. Chloroplast phylogeography Supplementary data associated with this article can be found, in of a temperate tree Pteroceltis tatarinowii (Ulmaceae) in China. J. Syst. Evol. 50, 325–333. the online version, at http://dx.doi.org/10.1016/j.ympev.2015.02. Liu, J.Q., Sun, Y.S., Ge, X.J., Gao, L.M., Qiu, Y.X., 2012. Phylogeographic studies of 016. plants in China: Advances in the past and directions in the future. J. Syst. Evol. 50, 267–275. Ni, J., Yu, G., Harrison, S.P., Prentice, I.C., 2010. Palaeovegetation in China during the References late Pleistocene: Biome reconstructions based on a global scheme of plant functional types. Palaeogeogr. Palaeoclim. Palaecol. 289, 44–61. Niu, X.Y., 2013. The Palaeo-Climate and Palaeo-Vegetation of XH-2 Core in Subei Bandelt, H.J., Forster, P., Röhl, A., 1999. Median-joining networks for inferring Basin Since the Mid-Pleistocene. MA Dissertation, Nanjing Normal University, intraspecific phylogenies. Mol. Biol. Evol. 16, 37–48. Nanjing. Chen, Z.D., Tatemi, S., 2001. Lardizabalaceae. In: Wu, C.-Y. (Ed.), Flora of China, vol. Pons, O., Petit, R.J., 1996. Measuring and testing genetic differentiation with ordered 6. Missouri Botanical Garden Press, St. Louis, pp. 440–454. versus unordered alleles. Genetics 144, 1237–1245. Chen, D., Zhang, X., Kang, H., Sun, X., Yin, S., Du, H., Yamanaka, N., Gapare, W., Wu, Posada, D., Crandall, K.A., 1998. Modeltest: testing the model of DNA substitution. H.X., Liu, C., 2012a. Phylogeography of Quercus variabilis based on chloroplast Bioinformatics 1, 817–818. DNA sequence in East Asia: multiple glacial refugia and mainland-migrated Qi, X.S., Chen, C., Comes, H.P., Sakaguchi, S., Liu, Y.H., Tanaka, N., Sakio, H., Qiu, Y.X., island populations. PloS One 7, e47268. 2012. Molecular data and ecological niche modelling reveal a highly dynamic Chen, Y., Compton, S.G., Liu, M., Chen, X.Y., 2012b. Fig trees at the northern limit of evolutionary history of the East Asian Tertiary relict Cercidiphyllum their range: the distribution of cryptic pollinators indicated multiple glacial (Cercidiphyllaceae). New Phytol. 196, 617–630. refugia. Mol. Ecol. 21, 1687–1701. Qian, H., Ricklefs, R.E., 2001. Diversity of temperate plants in East Asia – reply. Chou, Y.W., Thomas, P.I., Ge, X.J., LePage, B.A., Wang, C.N., 2011. Refugia and Nature 413, 130. phylogeography of Taiwania in East Asia. J. Biogeogr. 38, 1992–2005. Qiu, Y.X., Guan, B.C., Fu, C.X., Comes, H.P., 2009. Did glacials and/or interglacials Clark, P.U., Archer, D., Pollard, D., Blum, J.D., Rial, J.A., Brovkin, V., Mix, A.C., Pisias, promote allopatric incipient speciation in East Asian temperate plants? N.G., Roy, M., 2006. The Middle Pleistocene Transition: characteristics, Phylogeographic and coalescent analyses on refugial isolation and divergence mechanisms, and implications for long-term changes in atmospheric pCO . 2 in Dysosma versipellis. Mol. Phylogenet. Evol. 2009 (51), 281–293. Quaternary Sci. Rev. 25, 3150–3184. Qiu, Y.X., Fu, C.X., Comes, H.P., 2011. Plant molecular phylogeography in China and Cui, Z.J., Chen, Y.X., Zhang, W., Zhou, S.Z., Zhou, L.P., Zhang, M., Li, C.C., 2011. adjacent regions: tracing the genetic imprints of Quaternary climate and Research history, glacial chronology and origins of Quaternary glaciations in environment change in the world’s most diverse temperate flora. Mol. China. Quaternary Sci. 31, 749–764. Phylogenet. Evol. 59, 225–244. Davis, M.B., Shaw, R.G., 2001. Range shifts and adaptive responses to Quaternary Rogers, A.R., 1995. Genetic evidence for a Pleistocene population explosion. climate change. Science 5517, 673–679. Evolution 49, 608–615. Ding, L., Gan, X.N., He, S.P., Zhao, E.M., 2011. A phylogeographic, demographic and Rogers, A.R., Harpending, H., 1992. Population growth makes waves in the historical analysis of the short-tailed pit viper (Gloydius brevicaudus): evidence distribution of pairwise genetic differences. Mol. Biol. Evol. 9, 552–569. for early divergence and late expansion during the Pleistocene. Mol. Ecol. 20, Shafer, A., Cullingham, C.I., CÔTÉ, S.D., Coltman, D.W., 2010. Of glaciers and refugia: 1905–1922. a decade of study sheds new light on the phylogeography of northwestern Drummond, A.J., Rambaut, A., 2007. BEAST: Bayesian evolutionary analysis by North America. Mol. Ecol. 19, 4589–4621. sampling trees. BMC Evol. Biol. 7, 214. Shaw, J., Lickey, E.B., Schilling, E.E., Small, R.L., 2007. Comparison of whole Dutton, A., Lambeck, K., 2012. Ice volume and sea level during the last interglacial. chloroplast genome sequences to choose noncoding regions for phylogenetic Science 337, 216–219. studies in Angiosperms: the tortoise and the hare III. Am. J. Bot. 94, 275–288. Emerson, K.J., Merz, C.R., Catchen, J.M., Hohenlohe, P.A., Cresko, W.A., Bradshaw, Shi, Y., 1998. Evolution of the cryosphere in the Tibetan Plateau, China, and its W.E., Holzapfel, C.M., 2010. Resolving postglacial phylogeography using high- relationship with the global change in the Mid Quaternary. J. Glaciol. Geocryol. throughput sequencing. Proc. Natl. Acad. Sci. USA 107, 16196–16200. 20, 197–208. Excoffier, L., Laval, G., Schneider, S., 2006. Arlequin ver. 3.1: an Integrated Software Shi, M.M., Michalski, S.G., Welk, E., Chen, X.Y., Durka, W., 2014. Phylogeography of a for Population Genetic Data Analysis. University of Bern, Switzerland. Available widespread Asian subtropical tree: genetic east–west differentiation and at:
Tajima, F., 1989. Statistical method for testing the neutral mutation hypothesis by Yi, S., Saito, Y., Zhao, Q., Wang, P., 2003. Vegetation and climate changes in the DNA polymorphism. Genetics 123, 585–595. Changjiang (Yangtze River Delta), China, during the past 13,000 years inferred Tzedakis, P.C., Raynaud, D., Mcmanus, J.F., Berger, A., Brovkin, V., Kiefer, T., 2009. from pollen records. Quaternary Sci. Rev. 22, 1501–1519. Interglacial diversity. Nature Geosci. 2, 751–755. Ying, T.S., Zhang, Y.L., Boufford, D.E., 1993. The Endemic Genera of Seed Plants of Wang, H.W., Ge, S., 2006. Phylogeography of the endangered Cathaya argyrophylla China. Science Press, Beijing. (Pinaceae) inferred from sequence variation of mitochondrial and nuclear DNA. Yu, G., Chen, X., Ni, J., Cheddadi, R., et al., 2000. Palaeovegetation of China: a pollen Mol. Ecol. 15, 4109–4122. data-based synthesis for the mid-Holocene and Last Glacial Maximum. J. Wang, F., Li, D.-Z., Yang, J.-B., 2002. Molecular phylogeny of the Lardizabalaceae Biogeogr. 27, 635–664. based on TrnL–F sequences and combined chloroplast data. Acta Bot. Sinica 44, Zhan, A., Fu, J., 2011. Past and present: Phylogeography of the Bufo gargarizans 971–977. species complex inferred from multi-loci allele sequence and frequency data. Wang, H.F., Kirchoff, B.K., Qin, H.N., Zhu, Q.X., 2009a. Reproductive morphology of Mol. Phylogenet. Evol. 61, 136–148. Sargentodoxa cuneata (Lardizabalaceae) and its systematic implications. Plant Zhang, Z.Y., Wu, R., Wang, Q., Zhang, Z.R., Fan, D.M., López-Pujol, J., Li, D.Z., 2013. Syst. Evol. 280, 207–217. Comparative phylogeography of two sympatric beeches in subtropical China: Wang, J., Gao, P.X., Kang, M., Lowe, A.J., Huang, H.W., 2009b. Refugia within refugia: species-specific geographic mosaic of lineages. Ecol. Evol. 3, 4461–4472. the case study of a canopy tree (Eurycorymbus cavaleriei) in subtropical China. J. Zhao, S.Q., 1986. Physical Geography of China. Science Press, Beijing. Biogeogr. 36, 2156–2164. Zhou, T.H., Li, S., Qian, Z.Q., Su, H.L., Huang, Z.H., Guo, Z.G., Dai, P.F., Liu, Z.L., Zhao, Wu, C.Y., 1980. Vegetation of China. Science Press, Beijing. G.F., 2010. Strong phylogeographic pattern of cpDNA variation reveals multiple Xiong, D., Xiong, Y., Xi, Z., Mei, J., 2007. High-yield cultivation technique of Akebia glacial refugia for Saruma henryi Oliv. (Aristolochiaceae), an endangered herb trifoliata (Thunb) Koidz. J. Hunan Agr. Univ. (Nat. Sci.) 33, 156–159. endemic to China. Mol. Phylogenet. Evol. 57, 176–188.