Does the Arcto-Tertiary Biogeographic Hypothesis Explain the Disjunct Distribution of Northern Hemisphere Herbaceous Plants? The Case of Meehania (Lamiaceae)

Despite considerable progress, many details regarding the evolution of the Arcto-Tertiary flora, including the timing, direction, and relative importance of migration routes in the evolution of woody and herbaceous taxa of the Northern Hemisphere, remain poorly understood. Meehania (Lamiaceae) comprises seven species and five subspecies of annual or perennial herbs, and is one of the few Lamiaceae genera known to have an exclusively disjunct distribution between eastern Asia and eastern North America. We analyzed the phylogeny and biogeographical history of Meehania to explore how the Arcto-Tertiary biogeographic hypothesis and two possible migration routes explain the disjunct distribution of Northern Hemisphere herbaceous plants. Parsimony and Bayesian inference were used for phylogenetic analyses based on five plastid sequences (rbcL, rps16, rpl32-trnH, psbA-trnH, and trnL-F) and two nuclear (ITS and ETS) gene regions. Divergence times and biogeographic inferences were performed using Bayesian methods as implemented in BEAST and S-DIVA, respectively. Analyses including 11 of the 12 known Meehania taxa revealed incongruence between the chloroplast and nuclear trees, particularly in the positions of Glechoma and Meehania cordata, possibly indicating allopolyploidy with chloroplast capture in the late Miocene. Based on nrDNA, Meehania is monophyletic, and the North American species M. cordata is sister to a clade containing the eastern Asian species. The divergence time between the North American M. cordata and the eastern Asian species occurred about 9.81 Mya according to the Bayesian relaxed clock methods applied to the combined nuclear data. Biogeographic analyses suggest a primary role of the Arcto-Tertiary flora in the study taxa distribution, with a northeast Asian origin of Meehania. Our results suggest an Arcto-Tertiary origin of Meehania, with its present distribution most probably being a result of vicariance and southward migrations of populations during climatic oscillations in the middle Miocene with subsequent migration into eastern North America via the Bering land bridge in the late Miocene.


Introduction
The biogeographic history of intercontinental disjunctions between eastern Asia and eastern North America has long fascinated botanists and biogeographers [1][2][3], but until the inception of molecular phylogenetics and the accompanying advance of complex analytical approaches, these disjunctions were generally poorly understood. During the past two decades, however, the phylogenetic relationships between disjunct lineages, the timing of these disjunctions, and putative migration pathways for many disjunct taxa have been elucidated using molecular data and new analytical techniques [4][5][6]. Most of these studies have focused on woody plants, but several studies have examined the evolution of these disjunct patterns in terrestrial herbs [7][8][9][10][11].
The primary hypothesis put forth for explaining patterns of East Asian/eastern North American floristic disjunctions has been that a once continuous Arcto-Tertiary flora existed in the Northern Hemisphere during the late Cretaceous and Palaeogene that was fragmented by extinction due to global climatic cooling during the Neogene and Quaternary [3,[12][13][14]. However, the wide range of divergence times estimated from molecular dating among disjunct taxa between eastern Asia and North America suggests multiple and complex origins of the disjunctions in the Northern Hemisphere [15]. Based on 98 lineages with disjunct distributions between the two regions, Wen et al. [6] hypothesized that most of these lineages originated in eastern Asia and subsequently moved to North America, but also postulated that some have migrated in the opposite direction. At the same time, several groups present a distinct pattern, such as Triosteum L.  [7,16,17]. The Arcto-Tertiary flora once occupied wide areas of northern high latitudes in Cretaceous and early Paleogene time [18,19], and this vegetation subsequently migrated southward to middle latitudes in Eurasia and North America [20]. During such movements in space and time, many taxa became extinct or restricted to central and southern China and/or eastern/western North America. However, the Arcto-Tertiary biogeographic hypothesis alone cannot explain the disjunct distribution of many taxa because of plant migration during more recent times. Two migration routes, the Bering land bridge (BLB) and the North Atlantic land bridge (NALB), are crucial in interpreting Northern Hemisphere floristic disjunctions [21][22][23][24]. Paleontological and molecular data suggest that the BLB was used mostly by temperate taxa prior to the late Miocene (<10 Mya) [6,13,15], while the NALB has been viewed as a crucial route for the spread of subtropical and tropical taxa in the early Paleogene [13,23,25]. Recently, the transoceanic long distance dispersal (LDD) has been proposed for taxa for which no land migration route existed at the time of migration, e.g. Kelloggia Torrey ex Benth. & J. D. Hooker of Rubiaceae [9] and Leibnitzia Cass. of Asteraceae [26].
Meehania is a small genus of annual and perennial herbaceous plants consisting of seven species and five subspecies [27]. Meehania has an unevenly disjunct distribution between eastern Asia (11 taxa) and eastern North America (1 taxon; Fig. 1). Perhaps in part due to its disjunct distribution, Meehania species were previously assigned to distant genera such as Dracocephalum L., Cedronella Moench, and Glechoma L. [27]. To date, the taxonomy of the genus, particularly the eastern Asian species, has only been assessed based on morphology. Morphological variation within Meehania is chiefly observed in inflorescences, calyx characters, and especially leaf morphology [27][28][29]. According to our field investigations and specimen examinations, however, leaf morphology is highly variable in different populations.
The genus Meehania is characterized by having stolons, cordate-ovate to lanceolate leaves, thyrsoid, terminal cymes, a pedunculate or sessile inflorescence with larger flowers (ca. 1-2.5 The Bayesian 50% majority-rule consensus tree of Meehania and closely related taxa inferred from analyses using (right) combined nuclear ribosomal DNA regions (ITS and ETS) and (left) combined chloroplast DNA regions (rbcL, rps16, trnL-F, rpl32-trnL and psbA-trnH). Numbers above the nodes are Bayesian posterior probabilities and below the nodes are bootstrap values obtained from MP analysis. cm long), a tubular calyx, a strongly 2-lipped and 5-lobed (3/2) corolla, and parallel antherthecae [27,31]. Cytological analyses based on two species of Meehania, M. urticifolia (Miq.) Makino and M. montis-koyae Ohwi, indicated that the genus is diploid, 2n = 18 [30]. Meehania, together with 12 other extant genera, belongs to subtribe Nepetinae, tribe Mentheae, but its systematic position within the subtribe is uncertain [31]. Although significant progress has been made in Lamiaceae phylogenetics at the tribal and generic levels [32][33][34][35][36][37][38], the genus Meehania has been underrepresented in molecular systematic studies. Thus far, only two molecular phylogenetic studies have included Meehania species [34,39]. In their study on tribe Mentheae, Meehania was included as a member of the subtribe Nepetinae by Drew and Sytsma [34]. They suggested that Meehania was polyphyletic because of the inclusion of the Eurasian genus Glechoma and Chinese endemic Heterolamium C. Y. Wu. However, their sampling was limited as their study only included two species of Meehania and one species of Glechoma. Furthermore, the voucher specimen for Heterolamium debile (Hemsl.) C. Y. Wu (Zhiduan, 960093) used in their study was subsequently found to be misidentified by the first author of this paper, and is in fact M. henryi (Hemsl.) Sun ex C. Y. Wu. Therefore, a comprehensive species sampling of both Meehania and Glechoma is vital for resolving relationships within and between the two genera.
Although Meehania is not especially species-rich compared with some other well-known Nepetoideae genera (e.g. Salvia L., Nepeta L.), its East Asian/North American disjunct distribution makes it well suitable for testing the hypothesis that Arctic latitudes in the Tertiary were a major center of origin for taxa currently occurring in East Asia and elsewhere in the North Hemisphere. It is noteworthy that of the~12 genera of subtribe Nepetinae, 3 possess analogous East Asian/North American disjunct distributions, suggesting common migration routes and similar evolutionary processes in these genera. Meehania species typically occur in temperate to subtropical forests in the Northern Hemisphere. In eastern Asia, M. urticifolia and M. montis-koyae are both restricted to northeastern China and Japan in temperate areas [27,28,40] [27,40]. In these southerly areas, Meehania taxa inhabit mesic sheltered microhabitats within coniferous or mixed evergreen broad-leaved forests in moist alpine areas and along valley streams. The perennial M. cordata (Nutt.) Britt. is endemic to eastern North America, and ranges from Southwest Pennsylvania in the north to North Carolina in the south, and is found as far west as southern Illinois. Few mints exhibiting a primarily East Asian-eastern North American disjunction pattern have been the primary focus of phylogenetic or biogeographic studies. Thus, Meehania offers an excellent opportunity to study biogeography and diversification of an East Asian/North American disjunct group distributed across the temperate and subtropical regions of two continents.
In order to test the hypothesis of an Arcto-Tertiary origin of Meehania and subsequent migration southward to south-central China and south-eastern North America, we collected accessions of Meehania throughout its range and employed DNA sequence data from both the nuclear ribosomal and chloroplast genomic regions to address the following specific questions: (1) Is Meehania monophyletic, and how is it related to Glechoma and other genera of Nepetinae? (2) When and where did Meehania evolve? and (3) what was the likely mechanism or route that facilitated the East Asian/eastern North American disjunction within the genus?

Ethics Statement
The authors have studied herbarium materials from the herbaria KUN and PE. No special permits were required for this study because all samples were collected by researchers with introduction letters of KIB (Kunming Institute of Botany, Chinese Academy of Sciences) in Kunming. Voucher specimens were deposited in the Herbarium, Kunming Institute of Botany, CAS (KUN). The plant materials did not involve endangered or protected species.

Taxon sampling
A total of 19 accessions belonging to 11 of the 12 currently recognized taxa of Meehania were included in this study (Table 1). Only M. pinfaensis (Levl.) Sun ex C. Y. Wu, a narrow endemic from Guizhou Province of southwestern China, was not sampled. Our sampling of Meehania covered the whole geographic range of the genus from southern and northern East Asia and eastern North America. All samples of Meehania in this study were wild collected and dried with silica-gel except for two accessions of M. urticifolia obtained from herbarium specimens ( Table 1). As recent phylogenetic studies of Mentheae show that Glechoma is the closest relative to Meehania [34,39], 10 accessions of Glechoma were included in this study (Table 1). Sequences of two Meehania and five Glechoma accessions from GenBank were also included in our analyses (S1 Appendix).
In addition to the taxon sampling above, we also sampled across the Nepetoideae for our divergence time analyses (see below). Voucher information and GenBank accession numbers for all specimens used in this study are listed in Table 1, as well as S1 Appendix.

DNA extractions, amplification, and sequencing
Total genomic DNA was isolated from silica gel-dried leaf material using a Universal Genomic DNA Extraction Kit (Takara, Dalian, China). Five chloroplast (rbcL; the rps16 intron; the trnL-F region; the rpl32-trnL and psbA-trnH intergenic spacers) and two nuclear ribosomal regions (ITS and ETS) were selected for phylogenetic inference. Primers used for amplification and sequencing were Z1 and 1204R for rbcL [41], F and 2R for the rps16 intron [42], and tabc and tabf [43] for the trnL-F region. The rpl32-trnL and psbA-trnH spacers were amplified using the primers as described by Shaw et al. [44] and Sang et al. [45], respectively. ITS was amplified and sequenced using the primers ITS1 and ITS4 [46], and ETS was amplified and sequenced as described in Drew and Sytsma [39]. Amplified DNA samples were analyzed by electrophoresis on 1.4% agarose gel, run in a 0.5 × TBE buffer and detected by ethidium bromide staining. The PCR products were then purified using a QiaQuick gel extraction kit (Qiagen, Inc., Valencia, California, USA) and directly sequenced in both directions using the amplification primers on an the ABI 3730 automated sequencer (Applied Biosystems, Forster City, California, USA).

Sequence alignment and phylogenetic analyses
DNA Baser v.3 (http://www.DnaBaser.com) was used to evaluate the chromatograms for base confirmation and to edit contiguous sequences. Multiple-sequence alignment was performed by MAFFT v.6 [47], using the default alignment parameters followed by manual adjustment in Se-Al v2.0a11 (http://tree.bio.ed.ac.uk/software/seal/), and gaps were treated as missing data.
Phylogenetic trees were constructed using maximum-parsimony (MP) and Bayesian inference (BI). The MP analyses were conducted using PAUPÃ version 4.0b10 [48]. All characters were weighted equally and unordered. Most parsimonious trees were searched with a heuristic algorithm comprising tree bisection-reconnection, branch swapping, MULPARS, and the Table 1. List of species/taxa with voucher information (Herbarium), location, GenBank accession for sequences of species used in this study.

Taxa
Voucher ITS ETS trnL-F rpl32-trnL psbA-trnH rbcL rps16  alternative character state. A strict consensus tree was constructed from the most parsimonious trees. Bootstrap analyses (BP; 1000 pseudoreplicates) were conducted to examine the relative level of support for individual clades on the cladograms of each search [49]. Nucleotide substitution model parameters were determined for cpDNA and nrDNA data sets using MrModeltest version 2.3 [50,51]. Bayesian inference was conducted using MrBayes version 3.2.1 [38,52] with the model parameters determined from MrModeltest. For the chloroplast DNA partitions MrModeltest suggested the K81uf+ G (rps16, psbA-trnH and trnL-F) and TVM+ G (rbcL and rpl32-trnL spacer) models. For the nrDNA partitions, MrModeltest suggested the TVM+ G model for ETS and GTR+ I+ G for ITS. The Markov chain Monte Carlo (MCMC) algorithm was run for 3,000,000 generations with one cold and three heated chains, starting from random trees and sampling one out of every 300 generations. Runs were repeated twice to test the convergence of the results. The burn-in and convergence diagnostics were graphically assessed using AWTY [53]. After discarding the trees saved prior to the burnin point (ca. 15%), the remaining trees were imported into PAUP and a 50% majority-rule consensus tree was produced to obtain posterior probabilities (PP) of the clades. The incongruence length difference (ILD) test [54] was used to evaluate congruence between the chloroplast and the nuclear data sets. For all ILD tests, 100 replications were performed using PAUPÃ. As the ILD test (P < 0.01) suggested incongruence between the two data sets, and the topologies also exhibited discordance, we performed separate analyses for the cpDNA and the nrDNA data.

Divergence time estimation
For our divergence time estimation, we analyzed the Meehania clade within a broad phylogenetic framework of Lamiaceae to enable multiple fossil calibrations. We included 79 taxa from Nepetoideae in our nrDNA dataset and 74 Nepetoideae taxa for the cpDNA dataset, of which 59 were obtained from GenBank (S1 Appendix). Eriophyton wallichii Benth. from the Lamioideae served as an outgroup.
Like most plant groups, the fossil record of Lamiaceae is fairly sparse [31], but there are several described fossils that are useful for calibration points. Hexacolpate pollen is a synapomorphy for subfamily Nepetoideae [31], but is otherwise very rare within angiosperms. Kar [55]  identified a middle Eocene hexacolpate pollen sample as Ocimum L., which is within the Ocimeae tribe of Nepetoideae. However, based upon the comments of Harley et al. [31], we followed the methodology employed by Drew and Sytsma [34] and placed the fossil calibration at the crown of Nepetoideae as opposed to elsewhere (crown of the Ocimeae). Following the procedure of Drew and Sytsma [34], for both the nrDNA and cpDNA datasets the Nepetoideae crown was constrained with a lognormal prior having an offset of 49 million years (Mya), a mean of 2.6, and a standard deviation (SD) of 0.5. In both datasets we also constrained the most recent common ancestor of Melissa L. and Lepechinia Willd. with a log-normal distribution having an offset of 28.4 Mya, a mean of 1.5, and a SD of 0.5. The offset was based on a fossil fruit of Melissa from the early-middle Oligocene [56,57]. Additionally, Lepechinia and Melissa were constrained to be monophyletic in both the nrDNA and cpDNA analyses. To prevent the root of the tree from "running away" [58], the root of both the nrDNA and cpDNA trees was constrained using a uniform prior distribution with a minimum of 49 Mya and a maximum of 84 Mya. The maximum age corresponded to the upper age estimate (from the 95% HPD) obtained for the family Lamiaceae in Drew and Sytsma [34]. Since the oldest crown date for the order Lamiales is 107 Mya [34,59], and the Lamiaceae is nested deeply within the Lamiales, the 84 Mya maximum age for Lamiaceae used here is conservative. Bayesian dating based on a relaxed-clock model [60] was used to estimate the divergence times of the main clades in Meehania using the program BEAST version 1.8.0 [61]. BEAST employs a Bayesian MCMC approach to co-estimate topology, substitution rates and node ages [62]. Based on the results from Modeltest, the nrDNA analyses were performed using the GTR model of nucleotide substitution with a G and invariant sites distribution with six rate categories, while for the cpDNA data the TVM + G model was employed. The tree prior model (Yule) was implemented in the analysis, with rate variation across branches assumed to be uncorrelated and lognormally distributed [60]. Posterior distributions of parameters were approximated using two independent MCMC analyses of 30,000,000 generations (sampling once every 5000 generations). Samples from the two chains, which yielded similar results, were combined after a 10% burn-in for each. Convergence of the chains was checked using the program Tracer 1.5 [63], and the effective sample size (ESS) was well over 200 for all categories.

Biogeographic analyses
Analysis of potential ancestral distribution areas of clades and taxa in Meehania was conducted using RASP 2.1b [64], which implements the S-DIVA (statistical dispersal-vicariance analysis) method [65]. The input file for RASP consisted of the 10,800 post-burn-in trees from our nrDNA BEAST analyses. Three areas of endemism were defined for the biogeographical analysis based on the extant distribution of the genus and the geological history: A, northeastern Asia; B, southeastern Asia; C, eastern North America. Because there were no species in our studied taxa distributed in more than two areas, the maximum range size was constrained to 2 in our analyses.

Phylogenetic analyses
The combined nrDNA data matrix had 1144 characters, 519 of which were variable and 339 were potentially parsimony-informative. The parsimony strict consensus tree was largely congruent with the Bayesian consensus tree, especially concerning the backbone of the Meehania phylogeny. The Bayesian consensus tree with PP and BP values is shown in Fig. 1 (right). The combined chloroplast DNA (rbcL, rps16, trnL-F, rpl32-trnL and psbA-trnH) matrix consisted 4727 of characters, of which 914 were variable and 426 potentially parsimony-informative.
Topologies from the parsimony strict consensus tree and the Bayesian tree are largely congruent, and the Bayesian tree with PP value and BP support is shown in Fig. 1 (left).
Phylogenetic analysis based on the nrDNA data supported the monophyly of Meehania (Fig. 1). In the nrDNA tree, all Glechoma taxa formed a clade sister to a clade of Meehania species with strong support (Fig. 1, BP = 100, PP = 1.00). By contrast, in the cpDNA tree, Glechoma was nested within (instead of sister to) the Meehania clade, and was sister to the southeastern Asian Meehania clade, but this relationship received weak Bayesian support (PP = 0.67) and no parsimony support (Fig. 1).
Within Meehania, four lineages were well recognized in both the nuclear and chloroplast datasets: M. cordata (North America), M. montis-koyae (Japan and East China), the M. urticifolia (Northeast Asia), and a clade including the remaining species from southeastern Asia. The phylogenies resulting from the cpDNA analysis showed that M. montis-koyae diverged first, whereas in the nuclear data analysis, M. cordata was the first-diverging lineage. Both nuclear and chloroplast results indicated phylogenetic relationships among M. henryi, M. fargesii, and M. faberi are uncertain.

Biogeographic analysis
The chronogram and results of divergence-time estimation based on the nrDNA are shown in Fig. 2. The divergence age between Meehania and its sister Glechoma was estimated at 11.88 Mya with 95% highest posterior density (HPD) of 8.40-16.10 Mya (node 1, Fig. 2). The crown age of Meehania (node 2, Fig. 2), indicating the disjunction of Meehania between eastern Asia and North America, was estimated at 9.81 Mya in the Miocene (95% HPD 6.70-13.07 Mya). The split between the southeastern Asian Meehania lineage from its northern relatives (node 3, Fig. 2) was estimated at 6.12 Mya (95%HPD: 4.17-8.67 Mya). Divergence time estimates based on the cpDNA generated very similar divergence time as those from nrDNA. The crown age of Meehania (including Glechoma) was estimated to be 11.7 Mya (95%HPD: 7.69-16.72; S1 Fig.). The disjunction between eastern North American M. cordata and eastern Asian species was estimated to be 7.58 Mya (95%HPD: 4.90-10.86; S1 Fig.).
In Fig. 3 we illustrate the results obtained from S-DIVA, as well as migration or dispersal routes. The results of the biogeographic inference indicated that the crown node of Meehania unequivocally originated in the northern part of eastern Asia. Following the crown divergence, the genus was found to have had two diversification routes: one is an early split from northeastern Asia to eastern North America between M. cordata and the remaining Meehania species; another is a north to south migration within eastern Asia (Fig. 3).

A reticulate evolutionary history of Meehania-Glechoma with chloroplast capture
The chloroplast and nuclear phylogenetic analyses produced conflicting results with respect to generic relationships in the subtribe Nepetinae (Fig. 1). The most striking difference between the two topologies is in the position of Glechoma and Meehania cordata. In the chloroplast DNA tree, species of Glechoma formed a well-supported clade embedded within Meehania ( Fig. 1; BP = 78, PP = 1.0), and sister to the south clade ( Fig. 1; PP = 0.66), the pattern found also by Drew and Sytsma [34,39] using chloroplast data and limited sampling of these two genera. In contrast, the nuclear topology clustered all members of Meehania as a single moderately-supported clade ( Fig. 1; BP = 56, PP = 0.98) and separated the Glechoma clade from Meehania with high support (Fig. 1; BP = 100, PP = 1.0). Discordance between nuclear and cytoplasmic data is common in plants [66][67][68][69]. One possible explanation for the conflicts has invoked introgression of the cytoplasmic genome from one species into the nuclear background of another (or vice versa) by interspecific hybridization [67,70], in which case the incongruent trees represent the different histories of cp-and nrDNA. Another possible cause is intra-individual polymorphism of nrDNA, which may arise through incomplete concerted evolution, and can cause paralogy problems or incomplete lineage sorting of nrDNA [71].

Phylogenetic relationships
Based on nrDNA results, two well-supported lineages were recognized within Meehania: one clade consists of the single species from eastern North America and the other contains all eastern Asian taxa (Fig. 1). Within the eastern Asian group, the geographically isolated M. montiskoyae is sister to the remaining species. Meehania montis-koyae is endemic to Japan and known only from the type locality in Mt. Koya in Kii Peninsula, Wakayama Prefecture. A suite of morphological characters found in M. montis-koyae, such as an erect and herbaceous habit, a height of 10-20 cm, abaxial leaves purple, a violet tubular calyx, and an arrangement of flowers in axillary pairs are quite unique within Meehania. Recently, Xia and Li [83] reported that M. montis-koyae is also found in eastern China and occurs on slopes within or at the edge of mixed forests. This plant was previously unknown from China and bridges the two distribution areas between China and Japan. The M. montis-koyae individual from China is closely related to the two Japanese individuals as inferred by our molecular data with high support ( Fig. 1; BP = 100, PP = 1.0). The current disjunction of M. montis-koyae between eastern China and Japan might be remant populations left over from a previously existing continuous distribution.
Except for Meehania montis-koyae and M. urticifolia, all the species from southeastern Asia form a well-supported south clade ( Fig. 1; BP = 80, PP = 1.0). Phylogenetic relationships of the three species complexes among the south clade remained unresolved (Fig. 1), possibly due to the recent evolutionary radiation of this group. However, taxa from this clade exhibit a wide range of morphological and ecological variations. Meehania faberi is a distinct species based on its annual life history, morphological traits such as ovate and fleshy leaves and short inflorescences, and a geographically isolated distribution [27]. The two geographically widespread species complexes, Meehania henryi and M. fargesii, were found to be polyphyletic (Fig. 1). The Meehania henryi complex is endemic to a small area of Central China and is characterized by an erect habit, a height of ca. 30-60 cm, large leaves, a narrowly tubular calyx, and verticillasters in terminal and lateral racemes [27,40]. The Meehania fargesii complex is characterized by having slender stems, a prostrate or stoloniferous habit, a height of 10-20 cm, a tubular calyx, and 2-flowered verticillasters inserted in the leaf axils of the upper 2 or 3 leaf pairs of the stem [27,40]. Subtle differences in verticillaster flower number, stem branching pattern and leaf shape were used previously to delimit subspecies within the complex [27]. Ecologically, the M. henryi complex is distributed in evergreen broad-leaved and mixed forests from 300-700 m in elevation, whereas the M. fargesii complex is distributed from temperate mixed forests to coniferous forests at a higher elevation from 700 to 3500 m.

Historical biogeography and divergence times
Glechoma, the sister group of Meehania, occurs in north temperate areas in Eurasia, and the basal lineages of Meehania (M. montis-koyae and M. urticifolia) are also largely restricted to northeastern Asia (i.e., Japan, East China, and South Korea) [27,28], making the high latitude area of Eurasia a plausible ancestral area for Meehania (Fig. 3). Ancestral area reconstruction with RASP based on our nrDNA phylogeny supported this view, suggesting a Meehania origin in the high latitude area of Eurasia, especially northeastern Asia (Fig. 3). This evidence agrees well with the Arcto-Tertiary origin hypotheses, which has been extensively documented [18,84,85]. Subsequently, the decrease of annual mean temperature at northern latitudes provided opportunities for biota dispersal and subdivision [86]. The present distribution of Meehania in eastern North America and northeastern and southeastern Asia could result from vicariance of south-migrating populations during climatic oscillation and further fragmentation and dispersal of these populations. This inference is robustly supported by our molecular phylogenetic results, viz. a sister relationship between North American M. cordata and the clade of East Asian Meehania (the latter comprising the two subclades within this area; Fig. 1). Similar cases are found in Astilbe Buch.-Ham. ex D. Don, Cedrus Trew, Maianthemum Web. and Triosteum L. in which the southeastern Asian species were found to have their origin in Arcto-Tertiary geofloras [7,17,87,88]. Zhu et al. [7] suggested Astilbe had its origin in Japan and subsequently migrated independently to eastern North America, continental Asia, and even to southeastern Asian islands. Based on fossils and molecular data, Qiao et al. [88] suggested an origin of Cedrus in high latitudes of Eurasia, and its present distribution in the Mediterranean and Himalayas could result from vicariance of a southward migration during climatic oscillations in the Tertiary.
The estimated divergence times between the Meehania lineages from isolated regions completely overlap the timing of Miocene cooling and drying. In the Miocene, a significant global cooling transition occurred at approximately 15-10 Mya [89][90][91]. This cooling event was proposed to cause southward invasions and displacements of organisms [92]. As a result, four Meehania species occur today in the southernmost areas of eastern Asia (Fig. 3). We estimated the divergence of the southern clade (between the northern M. urticifolia and other southern Asian taxa) at 4.17-8.67 Mya in the late Miocene. Another Miocene climate change emphasized by Savage [92] caused enhanced aridity at middle latitudes of the Northern Hemisphere. In the interior of Eurasia, a drying event occurred at about 8-7 Mya [93,94] that may have caused isolation between Meehania in northern and southern East Asia. Extant M. urticifolia and M. montis-koyae show preferences to cool and moist habitats [27,95], and are probably relicts that previously inhabited northern regions. This distribution pattern has also been reported for other taxa, such as Parthenocissus Planch. [96], Mitchella L. [8] and Astilbe Buch.-Ham. ex D. Don [7].
The ancestor of eastern North American Meehania might have reached North America in the late Miocene, which is supported by our estimation of ca. 9.81 Mya for the divergence between the North American M. cordata and the East Asian clade (Fig. 2). The North Atlantic land bridge, which largely contributed to the dispersal of more tropical elements, ceased to exist in the middle Miocene [13], and was apparently less suitable for Meehania interchange. We favor a hypothesis based on a migration scenario across the Bering land bridge in the late Miocene. North America and Asia were repeatedly connected via the Bering Bridge, with biotic interchange moderated mainly by climatic factors [97]. The Bering land bridge supported exchanges of temperate floras [3], but was ultimately disrupted by a sharp decrease in average temperatures from the Oligocene to the present [91]. In the late Miocene and Pliocene, the colder climate restricted Beringian interchange to mostly cold-adapted species. Decreasing temperatures could have prohibited subsequent interchange of warm adapted taxa, including Meehania, between eastern Asia and eastern North America.
Meehania, like other taxa from tribe Mentheae, possess mericarps for dispersal. The dispersal ability of these nutlets is usually limited (reviewed by [56,98]), and long distance dispersal between Asia and North America in Meehania is highly unlikely. Consequently, as a result of geographic and ecological isolation, diverged Meehania lineages likely formed within each aforementioned isolated region after these climatic change events. These results suggest that vicariance played an important role in the evolution of herbaceous plants between eastern Asia and North America.

Conclusions
Two important conclusions stem from this study. First, we show that Arctic latitudes were a major center of origin for taxa currently occurring in East Asia and elsewhere in the North Hemisphere. Secondly, the current disjunct distribution of some herbs with a putative Arcto-Tertiary origin is probably a result of vicariance and subsequent southward migration of populations during climatic oscillations in the middle Miocene with subsequent migration into eastern North America via the Bering land bridge in the late Miocene.