Molecular Phylogenetics and Evolution 68 (2013) 443–460

Contents lists available at SciVerse ScienceDirect

Molecular Phylogenetics and Evolution

journal homepage: www.elsevier.com/locate/ympev

Evolutionary events in Lilium (including Nomocharis, Liliaceae) are temporally correlated with orogenies of the Q–T plateau and the Hengduan Mountains ⇑ Yun-Dong Gao a,b, AJ Harris c, Song-Dong Zhou a, Xing-Jin He a, a Key Laboratory of Bio-Resources and Eco-Environment of Ministry of Education, College of Life Science, Sichuan University, Chengdu 610065, China b Chengdu Institute of Biology, Chinese Academy of Sciences, Chengdu 610041, China c Department of Botany, Oklahoma State University, Oklahoma 74078-3013, USA article info abstract

Article history: The Hengduan Mountains (H-D Mountains) in China flank the eastern edge of the Qinghai–Tibet Plateau Received 21 July 2012 (Q–T Plateau) and are a center of great temperate plant diversity. The geological history and complex Revised 24 April 2013 topography of these mountains may have prompted the in situ evolution of many diverse and narrowly Accepted 26 April 2013 endemic species. Despite the importance of the H-D Mountains to biodiversity, many uncertainties Available online 9 May 2013 remain regarding the timing and tempo of their uplift. One hypothesis is that the Q–T Plateau underwent a final, rapid phase of uplift 8–7 million years ago (Mya) and that the H-D Mountains orogeny was a sep- Keywords: arate event occurring 4–3 Mya. To evaluate this hypothesis, we performed phylogenetic, biogeographic, Hengduan Mountains divergence time dating, and diversification rate analyses of the horticulturally important genus Lilium, Lilium–Nomocharis complex Intercontinental dispersal including Nomocharis. The Lilium–Nomocharis complex is distributed throughout the temperate Northern Diversification rate Hemisphere but is most diverse within the H-D Mountains and Q–T Plateau. Our matK and ITS phyloge- Bayesian Binary Method (BBM) nies support previous studies showing that Nomocharis is nested within Lilium. However, we detected Reconstruct Ancestral State in Phylogenies incongruence between the two gene trees which may result from hybridization. Dating analyses per- (RASP) formed using the ITS dataset showed that the evolution of major lineages within Lilium–Nomocharis Biodiversity hotspot may be temporally coincident with Q–T Plateau uplift occurring 8–7 Mya and H-D Mountains uplift approximately 4–3 Mya. Our analyses of diversification times and rates among Lilium–Nomocharis clades are less conclusive. However, these do suggest high extinction rates among H-D Mountains lineages. Ó 2013 Elsevier Inc. All rights reserved.

1. Introduction mountains and deep valleys (Li et al., 1995), which are thought to have profoundly accelerated the diversification of plant species The Qinghai–Tibetan Plateau (Q–T Plateau) is bounded by asso- through local vicariance, secondary contact, and ecological specia- ciated mountain ranges, which reflect a complex history of uplift tion events (Liu et al., 2006). Thus, high species richness within the and quiescence in the region (Zhang et al., 2000; Pan et al., eastern Q–T plateau and adjacent areas may be strongly correlated 2012). The rich topography of the plateau and its broad-reaching with the region’s geomorphology. climatic influence have probably favored the evolution of the di- The Hengduan Mountain chain (H-D Mountains, Fig. 1; see also verse flora of eastern Asia; namely within the its mountain ranges http://hengduan.huh.harvard.edu/fieldnotes for further informa- but also within Japan and central China (Richardson et al., 2001a,b; tion) occupies the southeastern margin of Q–T Plateau and is Willis and Whittaker, 2002; Wharton et al., 2005; Liu et al., 2006). known as one of the world’s 25 or 34 biodiversity hotspots based Uplift of the plateau began approximately 40 million years ago on species richness and sensitivity to anthropogenic activities (Mya) in the Eocene (Chung et al., 1998). Recent evidence indicates (Myers et al., 2000; Wilson, 1992; http://www.biodiversityhot- that the southern margin reached its present elevation approxi- spots.org/xp/Hotspots; Richardson et al., 2001a,b). The region mately 15–22 Mya (Guo et al., 2002; Spicer et al., 2003) and that may be one of the most geologically complex in China (Pan, the whole plateau may have attained its present altitude by 1989) due to the extensive geomorphism that has occurred around 7–8 Mya (Harrison et al., 1992). Orogeny along the eastern margin the Sichuan Basin resulting from differences in crustal strength be- of the plateau since the early Miocene (22 Mya) has created high tween the basin and the surrounding, uplifted area (Royden et al., 2008). Evidence from molecular and floristic studies has implicated

⇑ Corresponding author. Fax: +86 028 85415006. the orogeny and complex topography of the H-D Mountains in E-mail address: [email protected] (X.-J. He). driving the high plant species richness observed in the region.

1055-7903/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.ympev.2013.04.026 444 Y.-D. Gao et al. / Molecular Phylogenetics and Evolution 68 (2013) 443–460

Fig. 1. Major geographical features of East Asia and the location of Hengduan Mountains (H-D Mountains).

For example, Li (1980) conducted a floristic survey the wetland speciation may provide insight into the timing of orogenic events plant genus Arisaema Mart. within the H-D Mountains and found (e.g., Antonelli et al., 2009; Picard et al., 2008; Stigall Rode and Lie- high diversity among both ‘‘primitive’’ and derived lineages. Thus, berman, 2005; Che et al., 2010). For example, a historical biogeo- Li (1980) hypothesized that multiple, divergent clades within graphic study of the family Rubiaceae was used to test the Arisaema underwent recent, independent diversifications within alternative hypotheses that the Andes Mountains rose continu- the region. Similarly, Wang et al. (2010) speculated that genetic ously or in stages of uplift and quiescence (Antonelli et al., 2009). differentiation among H-D Mountain populations of the aquatic Thus, it may be possible to test the competing hypotheses regard- herb Batrachium bungei (Steudel) L. is best explained by isolation ing the timing of the H-D Mountain orogeny using biological data. due to topographical complexity. Additional investigations using The genus Lilium, including Nomocharis, is an excellent model species-rich groups are needed to test the biological mechanisms system for testing the H-D Event hypothesis due to its geographic of these bursts of speciation. In particular, did uplift cause local distribution within the H-D Mountains and, more broadly, within vicariance, provide corridors for secondary contact, promote eco- Q–T Plateau region. Lilium is primarily distributed within the logical speciation, or all of these in concert? Northern Hemisphere and has three main ranges including the Further, despite frequent speculations about the role of the H-D Q–T Plateau in East Asia (including the Himalayas and the H-D Mountain orogeny in plant diversification, few studies have used Mountains), North America, and the Caucasus in Europe (Patterson the tools of evolutionary biology and historical biogeography to and Givnish, 2002; Liang, 1995). Of all these regions, the Q–T Pla- investigate the timing of these geological events. There are two teau harbors the greatest diversity of Lilium. Within Q–T Plateau conflicting hypotheses that seek to explain H-D Mountain uplift. region, there are 40 species of Lilium; 30 of these plus all species The most widely accepted of these suggests that the H-D Moun- of Nomocharis are endemic to the H-D Mountains (Lighty, 1968; tains rose in the latter stages of the eastern Q–T Plateau orogeny Baranova, 1969; Sealy, 1983; Liang, 1995; Liang and Tamura, and attained their present height around 7 Mya in the Miocene 2000). Further, a previous phylogenetic and biogeographic study (Harrison et al., 1992; Shi et al., 1998). More recently, Chen of Liliales revealed that the Himalayas were the geographic origin (1992, 1996) showed evidence from lithostratigraphy, biostratigra- of Lilium and that the genus was later dispersed into the rest of phy, magnetostratigraphy and geomorphology indicating that the Eurasia and North America (Patterson and Givnish, 2002). Despite uplift of the H-D Mountains occurred later in the Miocene, around the potential utility of Lilium and Nomocharis for better under- 3.4 Mya, following a period of relative geological stability, and he standing H-D Mountain orogeny, many questions about the origins called this the H-D Mountain Movement or Event. He proposed and intercontinental migrations of Lilium remain unanswered. that the H-D Movement should be recognized as an independent The inferred timing of evolution of Lilium also makes the genus phase in the Q–T Plateau orogeny (Chen, 1992, 1996). Previous a good model system. Previous studies show that Lilium probably authors have suggested that the timing and rate of biological evolved during the Miocene. Thus, eastern Asian clades may have Y.-D. Gao et al. / Molecular Phylogenetics and Evolution 68 (2013) 443–460 445 been affected by geological activity occurring within the H-D within or among sections (Comber, 1949; Lighty, 1960, and papers mountain region of the Q–T Plateau. Vinnersten and Bremer cited in). (2001) used rbcL sequences to estimate the divergence times of In this study, we used Lilium–Nomocharis as a model system for the major clades of Liliales and showed that the genus Lilium evaluating the timing and biological ramifications of the H-D evolved 10 Mya. Subsequent studies, in which dating was per- Mountain orogeny. However, we first addressed several remaining formed at the familial or ordinal levels, have supported the find- questions on evolutionary relationships and species boundaries ings of Vinnersten and Bremer (2001). Within Lilium, Ikinci_ within the complex by sampling taxa more completely than in pre- (2011) determined that members of the European section Lirioty- vious studies; particularly emphasizing H-D Mountain endemic pus diverged from the rest of the genus 9 Mya ago and that speci- species. Our objectives were to clarify the phylogenetic relation- ation within the section increased 6 Mya. Rates and times of ships between subgeneric groups within the Lilium–Nomocharis speciation events within Lilium outside of Europe have not previ- complex and to determine the causes of discordance between gene ously been investigated. However, the available data indicate that trees. Our resulting phylogenies were applied towards inferring the the Miocene epoch was important in the origin and diversification spatial and temporal history of Lilium–Nomocharis and towards of Lilium. Thus, we expect Lilium to show great utility in testing the evaultating alterative hypotheses about the timing and tempo of alternative hypotheses of early versus late uplift of the H-D uplift in the H-D Mountain region. Mountains. Lilium belongs to tribe Lilieae of Liliaceae (sensu stricto)(Takhta- 2. Material and methods jan, 1997) and contains about 120 species which are sister to Frit- illaria and are more distantly related to Cardiocrinum and 2.1. Plant materials and outgroup selection Notholirion (Hayashi and Kawano, 2000). Species of Lilium have tra- ditionally been subdivided into seven sections (Comber, 1949), on All species and populations newly sampled in this study were the basis of 15 morphological characters, including floral form, collected by the authors in the field except three species: Lilium growth habit, germination pattern, and bulb structure. The classi- longiflorum Thunberg var. scabrum Masamune was collected from fication scheme proposed by Comber has been more widely ac- cultivated material in Hainan Province, and samples of Lilium ros- cepted than others (e.g., Baker, 1871; Wilson, 1925; Haw, 1986; thonii and L. papiferum were obtained from herbarium specimens Baranova, 1988) and is therefore used in this work unless other- (SZ). For field collections, leaves were placed in silica gel while in wise stated. The relationship of Lilium to Nomocharis has been the field, allowed to dry, and stored at 80 °C until use. The vou- widely discussed since the latter was first described by Franchet cher specimens were deposited in the Herbarium of Sichuan Uni- (1889), and the most current circumscriptions accept seven species versity (SZ). within Nomocharis (Balfour, 1918; Evans, 1925; Sealy, 1950, 1983; We sampled 38 of 55 species of Lilium from China based on the Liang, 1984). Liang (1984) and Liang and Tamura (2000) mentioned classification presented in Flora of China (Liang and Tamura, 2000) the possibility of combining Lilium and Nomocharis, but opted to and these samples represent all four sections recognized in Flora retain two separate genera pending phylogenetic investigations. Reipublicae Popularis Sinicae (Liang, 1980) as well as the five of Since then, evidence from molecular phylogenetic studies shows Comber’s seven sections native to China (Comber, 1949). In total, that Nomocharis may be nested within Lilium and supports redefin- 101 species of Lilium were sampled to cover all of its major distri- ing Lilium to include Nomocharis (Nishikawa et al., 1999, 2001; butional areas (i.e., eastern Asia, North America, Europe and Cauca- Hayashi and Kawano, 2000; Lee et al., 2011; Gao et al., 2012). In sus), in which 45 species (including variants) were newly sampled most previous phylogenetic studies, sampling of Nomocharis was from Hengduan Mountains and its adjacent area of China. We also limited to one or two species and authors asserted that additional sampled six of seven species of Nomocharis accepted in Sealy sampling was needed before making taxonomic revisions (Nishik- (1950) and Flora of China (Liang and Tamura, 2000) as well as awa et al., 1999, 2001; Hayashi and Kawano, 2000; Rønsted et al., the newly described species, N. gongshanensis (Gao et al., 2012). 2005). Most recently, Gao et al. (2012) reconstructed the phylog- Thus, our sampling included all seven Nomocharis native to China eny of Lilium and Nomocharis using two molecular markers and with 13 accessions, and excluded only one species, N. synaptica including six species of Nomocharis. The results of their study sup- Sealy, which is endemic to India. Voucher information for all newly ported previous investigations showing the placement of Nomo- sampled individuals is given in Appendix A. charis within Lilium. Though Gao et al. (2012) have made progress on the status of Nomocharis within Lilium, their results 2.2. Molecular analysis show discordance between nuclear and chloroplast gene trees. Both types of markers support the recircumscription of Lilium to 2.2.1. DNA amplification include Nomocharis, but it is clear that details of the relationship Total DNA was isolated from silica gel-dried leaf tissue using a between the two genera require additional clarification. modification of the cetyltrimethyl-ammonium bromide (CTAB) One possible cause of incongruence between gene trees is protocol from Doyle and Doyle (1987). We used the universal hybridization. Hybridization in the Lilum-Nomocharis complex primers ITS4 and ITS5 (White et al., 1990) to amplify the entire has long been speculated. Sealy (1983) first mentioned hybridiza- internal transcribed spacer (nrITS) (including ITS1, 5.8S and ITS2). tion in Nomocharis when he said ‘‘In addition to the species there The matK region, consisting of approximately 1500 bp, was ampli- are plants which are, or may be, hybrids.’ Further, he noted that fied using the primers -19F (Molvary et al., 2000) and 1326R (Cué- divergent species of Nomocharis are prone to hybridization when noud et al., 2002). PCR for all primer pairs was performed with grown in close proximity (e.g., Nomocharis notabilis Sealy, Sealy, 50 ng genomic DNA in 20 ll reactions in a GeneAmp PCR System 1983). Gao et al. (2012) showed molecular and morphological evi- 9700 (Applied Biosystems, USA). The thermocycler protocol for dence for a hybrid species of Nomocharis, N. gongshanensis Y.D. Gao nrITS was as follows: the initial denaturation was 94 °C for 2 min et X.J. He. In contrast, there are no reports of naturally occurring and the final extension was 72 °C for 10 min. Thermocycling con- Lilium hybrids. However, there has been considerable speculation sisted of 35 repetitions of denaturation at 94 °C for 45 s, primer that they occur (e.g., Muratovic´ et al., 2010; Douglas et al., 2011). annealing at 55 °C for 45 s, and extension at 72 °C for 60 s. The Further, facilitated hybridization between lilies is very common thermocycler conditions for matK were as follows: 94 °C (2 min) in cultivation and many species can easily crossed with others 446 Y.-D. Gao et al. / Molecular Phylogenetics and Evolution 68 (2013) 443–460 initial denaturation, 35 cycles of 94 °C (60 s) denaturation, 52 °C 2.3. Dating the times of divergence (60 s) annealing, 72 °C (1.5 min) elongation and 72 °C (10 min) fi- nal elongation. In an ideal situation, divergence time dating should be con- Direct sequencing of ITS PCR products failed for Lilium longiflo- ducted using multiple fossil calibration points including both an- rum var. scabrum, L. sempervivodeum, and Cardiocrinum gigateum, cient and recent fossils (Sauquet et al., 2012). Thus, we began by so cloning was used to amplify in these species. First, the PCR prod- reviewing the fossil record of Liliales, with particular emphasis ucts were purified use TIANgel Midi Purification Kit (TIANGEN Bio- on Liliaceae and its sister family Smilacaceae (comprised of Smilax tech, Beijing, China), and then cloned with pGM-T cloning Kit L. including Heterosmilax Kunth). Bremer (2000) reviewed the Cre- following the manufacturer’s protocols (TIANGEN Biotech, Beijing, taceous lineages of monocots and observed that only a few system- China). Successful transformations were sent to Invitrogen Biotech atically informative monocot fossils pre-date the K-T boundary and Co. Ltd. (Shanghai, China) for plasmid extraction, purification, and none of these are attributable to Liliales. Lililes-like pollen is first sequencing. reported from the Aptian of the Cretaceous (125–112 Mya) and is For ITS, the same primers used for PCR amplification were used characterized by having a broad colpus, an elongate shape, and dis- for sequencing. The matK gene region was sequenced using the tinctive exine sculpturing, which is finer towards the ends and primers -19F, and trnK-2R, 390F and 1326R of Cuénoud et al. coarser along the broadest section (Doyle, 1973; Muller, 1981; Friis (2002). All sequencing was performed by Invitrogen Biotech Co. et al., 2011). However, since similar pollen occurs elsewhere Ltd. (Shanghai, China) using a ABI-3730XL DNA sequencer. among the monocotyledons, this type is not definitively attribut- able to Liliales (Doyle, 1973; Daghlian, 1981; Bremer, 2000; Friis et al., 2011). Thus, microfossils are not useful for divergence time 2.2.2. Sequence alignment and phylogenetic analyses dating within Liliales or subordinate genera. Leaves comprising In addition to the new sequences generated for this study, 55 the artificial genus (i.e., form genus) Haemanthophyllum Budantsev ITS sequences and 43 matK sequences of Lilium used in previous have been placed in Liliales and are reported from the Late Creta- studies were initially obtained from Genbank or directly from the ceous to Oligocene (70.6–23 Mya) (Erwin and Stockey, 1991). authors (Nishikawa et al., 1999; Rønsted et al., 2005; Ikinci_ et al., However, Haemanthophyllum includes heterogeneous cordate and 2006; Rešetnik et al., 2007; Nishikawa et al., unpublished). How- linear leaves, which may not belong to Liliales, and is in need of ever, we followed Harpke and Peterson (2008) and removed six taxonomic clarification (Erwin and Stockey, 1991). Erwin and Stoc- of the sequences obtained from GenBank (accession numbers: key (1991) described a rhizome and aerial stem, Soleredera rhizo- AB020439, AB020448, AB020450, AB020470, AM292432, morpha Erwin et Stockey, from the middle Eocene (48.6– AM292419) as possible pseduogenes because of variations within 40.4 Mya; Hills and Baadsgaard, 1967) of western North America one or more of the three highly conserved motifs of 5.8S. and placed it within Liliales, family incertae sedis. Stockey and The boundaries of ITS were determined by comparing the Wehr (1996) later indicated that the fossil was the earliest occur- aligned sequences with previously published Lilium sequences rence of Liliaceae in western North America, but did not clarify the (Nishikawa et al., 1999, 2001). The boundaries of matK were new interpretation. The oldest reliable fossil of Liliales is a leaf, decided by comparing with sequences of Lilium obtained from Ripogonum tasmanicum Conran, R. J. Carp. & G. J. Jord, that has re- Genbank. DNA sequences for both the matK and ITS regions were cently been documented from the early Eocene (51–52 Mya; Car- aligned initially using ClustalX (Thompson et al., 1997) and then penter et al., 2007) of Tasmania (Conran et al., 2009). This fossil by eye following the guidelines of Kelchner (2000) in MEGA4.0 is attributed to the family Ripogonaceae, which, along with Philesi- (Tamura et al., 2007). Thus, gaps were positioned to minimize aceae, is sister to Liliaceae + Smilacaceae (Stevens, 2001 ff.). Nota- nucleotide mismatches. Uncorrected pairwise nucleotide differ- bly, the fossil age falls within the uppermost end of the range of ences were determined using PAUP version 4.0b10 (Swofford, dates obtained by Vinnersten and Bremer (2001) for the diver- 2003). gence of Ripogonaceae (55.4–25.4 Mya) in their work on diver- Unweighted maximum parsimony analyses of the matK and ITS gences of major clades within Liliales. Within Liliaceae we found datasets were carried out using PAUP. For each analysis, maxi- no reliable records of macrofossils. Cockerell, 1922 described a fos- mum parsimony trees were sought using the heuristic search strat- silized reproductive axis with affinity to Ruscus L. (Liliaceae sensu egies of PAUP (with 1000 replicate analyses, random stepwise lato) from the Oligocene Florrisant Beds (34Mya; Evanoff et al., addition of taxa, TBR branch swapping, and setting the maximum 2001) of Colorado. However, the fossil has since been transferred number of trees to 50,000). Gaps were treated as missing data. to Fabaceae (Manchester, 2001) and Ruscus is now considered Bootstrap values were calculated from 1,000,000 replicate analyses within Asparagales (Stevens, 2001). using fast stepwise addition of taxa, and only those values compat- In contrast to Liliaceae, macrofossils of Smilacaceae are known ible with the majority-rule consensus tree were recorded. Bayesian from at least the Eocene and have been briefly reviewed by Ding analyses were conducted independently for matK and ITS using et al. (2011). Fossils of Smilacaceae are primarily leaves, which MrBayes version 3.1.2 (Ronquist and Huelsenbeck, 2003). For both strongly resemble the leaves of modern Smilax in their ovate to datasets, we used the GTR + G model of nucleotide substitution cordate shapes, acute apeces, and acrodromous primary veins with according to results from MrModeltest version 2.2 (Nylander, net-like higher order veins between them (Homes, 1993+; Ding 2004). Bayesian analyses were run from a random starting tree et al., 2011). Although Smilacaeae-like leaves may date to the Cre- using three hot chains (0.1 temperature increments) and one cold taceous (Bews, 1927; Berry, 1930; Daghlian, 1981; Ding et al., chain. Analyses were run for 10 million generations and the trees 2011), one of the most widely accepted earliest occurrence of the were saved to a file every 100 generations. The first 25% (or family is Smilax wilcoxensis Berry from the early Eocene (48.6– 25,000) trees were discarded as the ‘‘burn-in’’ and a majority-rule 55.8 Mya; U.S. Geological Survey, 2011) of Tennessee (Berry, consensus tree was calculated based upon the remaining trees. 1930; Dilcher and Lott, 2005; Pinilla and Renner, 2010). Tao et al. Incongruence between ITS and matK gene trees was assessed (2000) accept a fossil leaf reported from the Paleocene of Henan, using the incongruence length difference (ILD) test (Farris et al., China as older, but did not provide a description or images. The 1994) implemented as the homogeneity test in PAUP. The test limitation of using any leaf fossil of Smilacaceae in divergence time was performed with 100 partition-homogeneity test replicates, dating is that Smilax-like leaves occur elsewhere among the mono- using a heuristic search option with simple addition of taxa, TBR cots (Daghlian, 1981), although uncommonly (Homes, 1993+). De- branch swapping and MaxTrees set to 1000. spite some uncertainty regarding the affinities of Smilacaceae Y.-D. Gao et al. / Molecular Phylogenetics and Evolution 68 (2013) 443–460 447 leaves and given the absence of Liliaceae in the fossil record, we ated in TreeAnnotator using the product method (Drummond used Smilax wilcoxensis to constrain the divergence of Liliaceae and Rambaut, 2007). from Smilacaceae. We assumed that this was a safe fossil, although possibly a late one (see ‘safe but late’ in Sanquet et al., 2012), be- 2.4. Biogeography analysis cause at least some of the earlier, more questionable occurrences may also represent Smilacaceae. Initially we obtained ITS se- Seven distributional areas of the Lilium–Nomocharis group were quences for 25 Smilacaceae and 106 additional Liliaceae (data delimited based on McRae (1998) and Ikinci_ et al. (2006): the not shown) and added these to our existing alignment for diver- Himalayas and H-D Mountains (A), eastern Asia and Siberia (B), gence time dating. The additional sequences were aligned using western North America (C), eastern North America (D), the Cauca- Muscle (Edgar, 2004; http://www.ebi.ac.uk/Tools/msa/muscle/) sus (E), and Europe (F). Ancestral ranges, comprised of one or more and by eye. The ITS alignment was difficult and showed consider- distributional areas, were reconstructed using the recently devel- able sequence heterogeneity between Smilacaceae and Liliaceae oped Bayesian Binary Method (BBM) implemented in the software (data not shown). Preliminary phylogenetic reconstructions using Reconstruct Ancestral States in Phylogenies (RASP) ver. 2.0 (Yu this dataset produced unlikely results. Thus, we did not use it for et al., in preparation; http://www.mnh.scu.edu.cn/S-diva/). Using further analyses. this method, ranges are treated as binary characters, such that a Instead, we estimated dates for Lilium using a two-step proce- lineage may be either present or absent in an area at any point dure. Specifically, we obtained a date for the Lilieae tribe using along a branch. State changes between presence and absence are an rbcL phylogeny calibrated with Smilax wilcoxensis, and then governed by the rates of local extinction (presence to absence) we applied the date to calibrating the Lilieae crown node in the and dispersal (absence to presence), and these rates are modeled ITS and matK phylogenies. A requisite assumption of this approach within a Q matrix, which analogous to a substitution rate matrix. is that rbcL, ITS, and matK gene trees each support the monophyly For our analyses, we selected the F81 + G rate model. Under this of Lilieae and its phylogenetic position within Liliaceae. Previous model, the rates of local extinction and dispersal are estimated molecular phylogenetic studies based on the markers rbcL(Patter- during the analysis. The rates may be unequal to each other within son and Givnish, 2002; Shinwari and Shinwari, 2010) and matK each distributional area (F81 parameter) and unequal among dis- (Rønsted et al., 2005) have resolved the Lilieae tribe with high sup- tributional areas (G or gamma parameter). Reconstructions were port and showed that it is sister to Tulipeae (sensu Patterson and performed using consensus trees of matK and ITS, independently Givnish, 2002). In contrast, no prior studies using ITS have suffi- (due to phylogenetic incongruence), resulting from parsimony ciently broad or dense taxonomic sampling to assess the mono- and Bayesian analyses for a total of 4 BBM analyses. BBM extends phyly of Lilieae or its phylogenetic position. Thus, we constructed consensus trees to include two virtual outgroups each with the a maximum likelihood ITS phylogeny of Liliaceae (sensu Patterson same conditional probability distributions. We set the virtual out- and Givnish, 2002) using representative sequences of all genera in group distributions to A (Himalayas and H-D Mountains), which is the family except Clintonia Raf. and Medeola L., for which sequences consistent with a previous study showing this area as the origin of were unavailable. Two accessions of Trillium L. (Melanthiaceae, Lil- Lilieae (Patterson and Givnish, 2002). All BBM analyses were run iales) comprised the outgroup. Details of this analysis are pre- for 2 million generations using 9 hot Markov chains and 1 cold sented in Supplementary file 1. The results support the chain with temperature increments of 0.1. Widespread ancestral monophyly of Lilieae within Liliaceae (BS = 99%) and show that it ranges were constrained to a maximum of three distributional is sister to the Tulipeae clade. Thus, the rbcL, matK, and ITS gene areas. trees appear to be congruent in their support for the Lilieae tribe and its relationship to the rest of Liliaceae. 2.5. Analyses of diversification rates To perform the two-step dating of Lilium, we first obtained 36 rbcL sequences of species of Liliaceae and Smilacaceae from Gen- To better understand the tempo and timing of diversification bank (Appendix A), and aligned them using MEGA4.0 (Tamura within Hengduan clades of Lilium–Nomocharis compared to clades et al., 2007). Simultaneous phylogenetic reconstruction and dating in other geographic areas, we estimated lineage growth and speci- of the rbcL data set was performed using BEAST ver. 1.5.3 (Drum- ation and extinction rates through time. These analyses were per- mond and Rambaut, 2007) and employed a soft lower bound (see formed using the R package Laser (Rabosky, 2006; http://cran.r- Yang and Rannala, 2006; also see Ho and Phillips, 2009) to con- project.org/web/packages/laser/index.htm), which takes dated strain the divergence time between Liliaceae and Smilacaceae. This phylograms in Newick format as input. For our analyses, we used constraint was set using a lognormal distribution based on the age the rooted matK and ITS trees generated in BEAST as well as sub- of Smilax wilcoxensis (offset 48.6, mean 52 and the 95% high prob- trees within these. To obtain each subtree, we pruned the ITS or ability density (HPD) 48.6–65 Ma, which covered the whole Paleo- matK tree in Mesquite (Maddison and Maddison, 2011) so that only cene and the early Eocene). In the second step, the Lilieae crown terminals comprising an individual clade remained. We repeated node of the ITS and matK phylogenies (independently) was cali- this to produce subtrees for all of the clades indicated in Figs. 1 brated using a normal prior distribution on the root age with mean and 2 (section names for ITS, numbered lineages in matK). This re- and 95% HPD to fit those obtained using the rbcL data. sulted in 11 ITS subtrees and nine for matK. For each of the 22 trees All estimations of divergence times obtained using BEAST were and subtrees, we used Laser to read branch times (getBtimes), performed as follows unless otherwise specified: We assumed that which were requisite for computing lineage growth through time substitution rates varied according to the uncorrelated lognormal (plotLTT) and estimating speciation and extinction rates. Modeling (UCLN) model (Drummond et al., 2006) and that the complex of speciation and extinction rates was done using the Laser option GTR + I + G model best represented nucleotide substitution. For fitBOTHVAR, which allows the rates to vary through time. We used the distribution of divergence times, a pure birth branching pro- default settings for the fitBOTHVAR model parameters except that cess (Yule model) was chosen as a prior. We ran two independent we applied a smaller lower bound of 0.01 on the net diversification Markov chains for 50,000,000 generations each from a random rate. We plotted the results of the fitBOTHVAR model using the starting tree. The chains were sampled every 1000 generations, plotRate command. Finally, we inferred the timing of the most sig- but the first 20% of these were eliminated as burn-in. All log and nificant diversification rate changes within the ITS and matK trees tree files from independent, simultaneous runs were combined (not subtrees). We accomplished this in Laser by fitting the branch- using LogCombiner, and maximum credibility trees were gener- ing times to a three rate model, yule3rate, which returns the two 448 Y.-D. Gao et al. / Molecular Phylogenetics and Evolution 68 (2013) 443–460

Fig. 2. Phylogenetic tree resulting from a Bayesian analysis of ITS. Clade names based on Comber (1949) and Liang (1980). Distributional areas of clades indicated. Support values shown on braches; Bayesian posterior probabilities (PP) on left and parsimony bootstrap (BS) on right. Y.-D. Gao et al. / Molecular Phylogenetics and Evolution 68 (2013) 443–460 449 absolute times from a dated phylogeny that rate shifts have oc- L = 999, consistency index, CI = 0.4555, retention index, curred. We limited possible shift times to nodes (ints = NULL) be- RI = 0.7997; in matK: L = 216, consistency index, CI = 0.8194, reten- cause allowing times along internodes produced only negligibly tion index, RI = 0.9318). Bayesian and parsimony were generally different results in preliminary analyses. congruent for ITS and matK, independently, and therefore further discussion is based on the Bayesian trees. The Bayesian ITS tree is comprised of a backbone polytomy 2.6. Testing for incomplete lineage sorting composed of three major lineages; the Archelirion + Leucolirion lineage, the L. duchartrei lineage, and a large lineage of all other Lili- Based on the results of phylogenetic analyses of ITS and matK, a um–Nomocharis (Fig. 2). The former two lineages have an eastern plastid incomplete lineage sorting hypothesis was developed to ex- Asian distribution, while the latter occurs throughout the Northern plain the incongruence between the gene trees (Supplementary file Hemisphere. Within the Archelirion + Leucolirion and the large Lili- 2). To test this hypothesis we used a modification of the method um–Nomocharis lineages, there are two and five moderately to proposed by Sang and Zhong (2000). We considered that a node, highly supported crown clades, respectively (Fig. 2). In addition Y, represented the last shared ancestor (LSA) of taxon A and B, to these seven crown clades, we recognize three additional ones. which had incongruent placement between our two gene trees. In particular, the Lilium–Nomocharis lineage also contains the Since we expect that lineage sorting occurred within the plastid Lophophorum clade, which was supported by our parsimony, but genome, we assumed that Y represents A + B on the ITS tree, while not Bayesian, analyses of ITS (results not shown). Further, we rec- Y is a floating node (Pagel et al., 2004) on the matK tree represent- ognize the Lilium (BS = 86%, PP = 1.00) and Sinomartagon clades ing A + (B + all intervening groups) (Supplementary file 2). In the within the Lilium–Nomocharis lineage. The ITS results do not sup- case that B is a hybrid, we expect the age of Y on the ITS tree to port delimitation of these two clades from one another. Thus, our be the same as Y on the matK tree; i.e., the hybridization event pro- ITS data support 11 major crown clades: L. duchartrei, seven re- ducing B occurred at a single point in time (Supplementary file 2). solved clades within the Archelirion + Luecolirion and Lilium– In contrast, if incomplete lineage sorting best explain the incongru- Nomocharis lineages, and three additional clades accepted based ent position of B, we expect that node Y on the matK tree to be old- on support from the parsimony analysis and morphological syna- er than on the ITS tree, since the divergence of the plastid lineages pomorphies (see Baker, 1871; Wilson, 1925; Comber, 1949; Haw, must have occurred prior to the divergence of B from A. 1986; Baranova, 1988). To test our incomplete lineage sorting hypothesis, we per- Four of the major crown clades resolved in our analyses of ITS formed divergence time analyses in BEAST for each gene tree using are congruent with sections delimited by Comber (1949); Archeli- the same data sets, branch evolution parameters (i.e., Yule pure rion (PP = 1.00; BS = 98%), Martagon (PP = 1.00; BS = 95%), Lirioty- birth), and models of substitution as described in the above sec- pus (PP = 1.00; BS = 82%), and Pseudolirium (PP = 1.00; BS = 79%). tion, Dating the times of divergence. Taxon sets including all descen- Among these, Archelirion and Martagon are distributed in Asia (ex- dents of three nodes (Y1, Y2, Y3) for each tree, respectively, were cept that Lilium martagon has a wide distribution extending into established and monophyly of the sets was enforced. Analyses central and western Europe), Liriotypus is a well-delimited clade were run for 50 million generations with sampling and logging of European lilies, and Pseudolirium is endemic to North America. every 1000 generations. The resulting log files were imported into All Nomocharis species occur together in a crown clade which Tracer v1.4 (Rambaut and Drummond, 2007) and viewed with bur- also includes several species of Lilium endemic to the Q–T Plateau nin set to 20% or 10 million generations. After the appropriateness (the Nomocharis clade; BS = 63%, PP = 1.00). Within the Nomochar- of the burnin was established visually, the mean divergence time is clade, species of Nomocharis were resolved into two lineages and standard deviation (derived manually from the standard error which are consistent with the classifications of Sealy (1983) and output) were used to perform T-tests. For each node Y, we tested Liang (1984). However, the presence of Lilium species with the the hypothesis that Yn(matK) = Yn(ITS). clade indicate that Nomocharis is not a natural group. There is some To determine the value and sensitivity of our method of distin- support for a sister relationship between the Nomocharis and Liri- guishing between hybridization and incomplete lineage sorting, otypus clades (BS < 50%, PP = 1.00). we used the method to evaluate Yn(matK) = Yn(ITS) for an incongruent The matK dataset also resolved Lilium into three major lineages relationship for which there was strong, independent evidence of forming a backbone polytomy, but there was considerable discor- hybridization. This test followed the same procedures outlined dance in the composition of these lineages compared to the three above and was run concurrently with evaluations of Y1–Y3 (Supple- resolved in the ITS phylogeny (Fig. 3). One major lineage consists mentary file 2). of taxa endemic to Asia including all sampled Q–T Plateau Lilium as well as Nomocharis (PP = 1.00; BS = 57%). Another lineage is 3. Results comprised of North American crown clades (PP = 1.00; BS = 63%) and the third is a disjunctive lineage including European and Asian 3.1. Phylogenentic analysis lilies (PP = 1.00; BS = 86%). The matK gene tree is congruent with only two sections delimited by Comber (1949), sects. Liriotypus The final ITS dataset consisted of 123 accessions, including the (PP = 1.00; BS = 94%) and Pseudolirium (PP = 1.00; BS = 63%), which seven outgroup sequences (Appendix A). The total ITS sequence form monophyletic crown clades. In contrast to the ITS phylogeny, alignment with gaps was 662 bp in length and consisted of 344 the analysis of matK resulted in a polyphyletic Nomocharis but with variable sites (52%) and 226 potentially parsimony-informative low support for intervening nodes. characters (34%). The matK dataset comprised 100 accessions and 89 taxa. The sequence length varied from 1248 to 1266, of which 3.2. Divergence dating and biogeographic reconstructions 169 sites were variable and 94 sites (7.4%) were parsimony infor- mative. The ILD test gave a value of P = 0.01, indicating significant Divergence time analyses based on rbcL and the Smilax wilcox- incongruence between the nuclear and chloroplast markers. Thus ensis calibration point, resulted in a median divergence age for Lil- the nuclear and plastid makers were not combined for down- ieae of 16 Mya with a high probability density (HPD) of 9.10– stream analyses. 23.20 Mya (results not shown). Thus, a normal distribution with In both ITS and matK data, parsimony analyses resulted in the a standard deviation of 1 and centered about the median, maximum number of trees, 50,000, of equal length (in ITS: 16.15 Mya, was applied to the Lilieae root node for dating of the 450 Y.-D. Gao et al. / Molecular Phylogenetics and Evolution 68 (2013) 443–460

Fig. 3. Phylogenetic tree resulting from a Bayesian analysis of matK. Clade names based on Comber (1949) and Liang (1980). Distributional areas of clades indicated. Bayesian posterior probabilities (PP) on left and parsimony bootstrap (BS) on right. Clades showing significant incongruence with ITS tree highlighted in gray. Y.-D. Gao et al. / Molecular Phylogenetics and Evolution 68 (2013) 443–460 451

Fig. 4. Ultrametric chronograms showing divergence time dating and biogeographic results based on the ITS phylogeny. Scale bar at bottom indicating branch length of 2 Mya. Clades (i.e., Fig. 2) collapsed for clarity. Mean divergence age given on nodes. Bars on nodes indicate the 95% HPD for divergence ages. Pie charts show probabilities of ancestral area reconstructions, colors of pie slices defined in legend. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

ITS and matK phylogenies. Despite phylogenetic discordance be- Analyses of speciation and extinction rates for the ITS and matK tween the ITS and matK gene trees, estimates of divergence times trees both show general stability through time (Figs. 6 and 7). Spe- from both analyses showed that Lilium evolved approximately cifically, extinction rates have stayed approximately the same 13.6 Mya (Figs. 4 and 5). Many of the major clades within Lilium since the evolution of Lilium–Nomocharis and speciation rates have emerged around 6–8 Mya followed by a burst of speciation approx- declined slightly. Thus, results from both phylogenies predict that imately 4 Mya. net diversification has somewhat diminished over time. Among Our ancestral range reconstructions inferred from the ITS phy- subtrees, Hengduan and Himalayan clades appear more suscepti- logeny show high support (PP = 1.00) for an origin of Lilium in ble to high rates of extinction compared to lineages in other areas, the Q–T Plateau region (Fig. 4). If the ITS reconstruction results while eastern Asian clades show a general trend towards rapidly with the highest PP are assumed to be accurate, then seven dis- declining speciation rates (Figs. 6 and 7). persals occurred within the history of Lilium and can completely explain its modern distribution. Four of these dispersals were dis- 3.4. 4 Testing for incomplete lineage sorting persals out of the Q–T Plateau. In contrast, reconstructions based on the matK tree showed less resolution at many nodes, particu- The results of the incomplete lineage sorting versus hybridiza- larly deep ones, and suggest that more dispersals may have been tion test is presented in Supplementary Table S1. At all nodes required (Fig. 5). The most highly supported ancestral range of Lili- tested, we found discordance of divergence times between the um based on results from matK was widespread including the Q–T ITS and matK phylogenies. Thus, these analyses reject hybridiza- Plateau region and eastern Asia with PP support of 0.75 (Fig. 5). tion in favor of the lineage sorting hypothesis. However, the anal- Further, the matK reconstructions suggest that two independent yses also reject hybridization in a case where there is considerable local extinction events are primarily responsible for Q–T Plateau additional evidence to support it. endemic lineages (Fig. 5).

3.3. Analyses of diversification rates 4. Discussion

Lineage through time analyses for the ITS and matK phylogenies 4.1. Hybridization versus incomplete lineage sorting (ILS) predict relatively stable growth rates beginning approximately 11 Mya and 6 Mya after lineage birth (i.e., 16 Mya, see divergence Based on the ILD test, the incongruence between our nuclear time methods and results), respectively (Supplementary Figs. S1 and plastid phylogenies of Lilium–Nomocharis was significant. and S2). Fitting data from each tree to a three rate model predicted Observations of our phylogenies show that incongruence are pri- that the greatest diversification rate changes for ITS both occurred marily deep ones that affect the topological positions of major approximately 4 Mya (4.1257 and 4.1259 Mya). For matK, the clades, not positions of species within clades. Such discordances greatest diversification rate changes were very recent; 0.5287 between phylogenies based on nuclear and chloroplast markers and 0.4496 Mya. are generally caused by convergent evolution, lineage sorting, or 452 Y.-D. Gao et al. / Molecular Phylogenetics and Evolution 68 (2013) 443–460

Fig. 5. Ultrametric chronograms showing divergence time dating and biogeographic results based on the matK phylogeny. Scale bar at bottom indicating branch length of 2 Mya. Mean divergence age given on nodes. Bars on nodes indicate the 95% HPD for divergence ages. Pie charts show probabilities of ancestral area reconstructions, colors of pie slices defined in legend. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

ancient hybridization and introgression (Rieseberg and Soltis, L. tigrinum may be allotriploid (Stewart and Bamford, 1943; Moens, 1991; Mason-Gamer et al., 1995; Rieseberg et al., 1996; Wendel 1969), but molecular phylogenetic studies have not been used to and Doyle, 1998; Sang and Zhong, 2000; Comes and Abbott, verify that the genetic complement of the species is the product 2001; Linder and Rieseberg, 2004). Among these, hybridization of two distinct progenitors. While there is only scant evidence and incomplete lineage sorting have gained considerable attention for hybridization in Lilium–Nomocharis, intragenic hybrids are sus- and are known to confound phylogenetic analyses (Sang and pected to have played important roles in diversification of other Zhong, 2000; Maddison and Knowles, 2006; Joly et al., 2009). Liliaceous genera including Gagea (Peterson et al., 2009) and Ery- We expected that the ILD test might reveal significant incongru- thronium (Allen, 2001). Thus, hybridization in addition to the doc- ence between the two gene trees because there is at least one nat- umented case in N. gongshanensis may be responsible for gene tree urally occurring hybrid within the Lilium–Nomocharis complex. incongruence. This species, N. gongshanensis, was documented by Gao et al. Though the Sang and Zhong (2000) test produced results favor- (2012) as a morphologically distinct Nomocharis having two copies ing our lineage sorting hypothesis, we suspect that the incongru- of ITS having affinities to two different intrageneric groups. Specu- ence we observed was caused, instead, by ancient hybridization lations have been made about natural hybridization occurring else- for several reasons. First, incomplete lineage sorting is not well where among the traditional Lilium genus (Muratovic´ et al., 2010; documented in Liliaceae. Zarrei et al. (2009) found incongruence Douglas et al., 2011; T.J. Givnish, pers. comm., 2011), but none of between plastid and nuclear ITS phylogenies of the Gagea–Lloydia these have been directly documented. Cytological studies indicate complex (Tulipeae tribe, sister to Lilieae; Patterson and Givnish, Y.-D. Gao et al. / Molecular Phylogenetics and Evolution 68 (2013) 443–460 453

Fig. 7. Speciation and extinction rates of the matK phylogeny and major clades.

the matK phylogeny. The geographic partioning observed among matK clades may result from chloroplast capture, whereby the cytoplasm of one species is replaced by that of another species Fig. 6. Speciation and extinction rates analysis of the ITS phylogeny and major through hybridization or introgression (Fehrer et al., 2007). In par- clades. ticular, limited seed dispersal but ample pollen dispersal by pollin- ators such as hummingbird, bumblebee and butterfly (Skinner, 1988; Liu Changqiu, pers. comm., 2013) may result in geographi- 2002). They did not propose incomplete lineage sorting as an cally stable maternal lineages (i.e., cpDNA phylogeny) despite explanation for the discordance, but the possibility cannot be ex- introgression events. We found that geographically partitioned cluded based on the data they presented. Second, the coalescence clades were particularly common among the haplotypes of the of organelle DNA may be four times faster than in nuclear genes Sino-Himalayan-Japanese region, such as in Lineage VI and V (Moore, 1995). Some reports suggest that ITS evolves much faster (Fig. 3). This may indicate that unequal limitations to dispersal be- than chloroplast DNA, but the cases are few and have been recog- tween seeds and pollen in these lineages have been especially nized as special ones (Buckler and Holtsford, 1996; Sang et al., important in the diversification of eastern Asian lilies. 1997). Therefore, it is unlikely that lineage sorting of nuclear genes An additional reason why we cannot reject introgression is that had been completed before the divergences of the incongruent our Sang and Zhong (2000) test may be subject to Type I error. clades from their common ancestor, while polymorphisms in chlo- When applied to Nomocharis gongshanensis, which has been shown roplast genes were retained in their common ancestors. Thus, the to be a hybrid (see Gao et al., 2012), the hybridization hypothesis apparent rarity of incomplete lineage sorting in Liliaceae and the (i.e., the null, that divergence times are equal) was rejected (Sup- relative, rapid coalescence times of chloroplast genes may lend plementary Table S1). Sang and Zhong (2000) originally applied some support to a hybridization explanation for incongruence. their method to divergence times derived from molecular clock Further, lineage sorting is a stochastic process and would not be methods that produced singular results or absolute divergence expected to show the strong geographical partitioning observed in times. In contrast, we applied the method to results from BEAST, 454 Y.-D. Gao et al. / Molecular Phylogenetics and Evolution 68 (2013) 443–460 which consisted of high probability densities of possible diver- as distinct species. In the chloroplast phylogeny, Lilium duchartrei gence times at each node. Comparing mean ages obtained from and L. lankongense were not resolved as sister to one another the BEAST posterior distributions may have been too stringent of (Fig. 3). Thus, the matK data suggest an even more distant relation- a test. In fact, when we do not apply a test but, rather, consider ship between the two species. The distributions of these two H-D the median ages resulting from BEAST analyses of the ITS and matK Mountain are to the north (Lilium duchartrei) and south (Lilium datasets (Figs. 4 and 5), we see that the ancestor of Nomocharis II duchartrei) of the Daxue Mountains (ca. 28°3400N), which occur would have introgressed with Sinomartagon I around 5.8 Mya within the broader H-D range and are located between Sichuan (matK, Fig. 5), at approximately the same time that Nomocharis and Yunnan Provinces. This lineage arose early within the Lilium– split into two major clades (5.7 Mya, ITS, Fig. 4). Nomocharis complex; approximately 12.8 Mya based on ITS, and Aside from ILD and hybridization, undetected paralogy among the two species split about 3.2 Mya (Fig. 4). Meager diversity with- ITS sequences could also explain discordance between the ITS in this relatively old lineage may be explained by high extinction and matK gene trees. Although we excluded ITS pseduogenes based rates (Figs. 6 and 7). on observed variations within the conserved motifs of 5.8S (de- scribed in Section 2), we cannot rule out the presence of unde- 4.2.3. Lophophorum tected, functional ITS paralogs within our data set. This is Darkly-colored tepal bases and complicated nectary structures especially true because we did not clone ITS for all newly sampled led Balfour (1918) to treat of Lophophorum species within Nomo- species and cannot vouch for the orthology of sequences obtained charis, which possesses similar features. Despite superficial resem- from GenBank or other sources. However, it seems unlikely that blances, our ITS phylogeny shows that the Lophophorum clade is paralogy among ITS sequences explains the discordance with the only distantly related to Nomocharis. Thus, the shared morpholog- matK tree since the ITS phylogeny recapitulates morphological ical features with Nomocharis are homoplasies that have arisen at similarities. Future studies may use more extensive cloning of ITS least twice in the Lilium–Nomocharis complex. or other nuclear genes to better detect and understand discordance In addition to their shared morphology, Lophophorum and between the organellar and chromosomal genomes in Lilium– Nomocharis also have sympatric ranges within alpine meadows of Nomocharis. the H-D Mountains. Convergent morphological and niche evolution may have occurred within these two divergent groups due to envi- 4.2. The phylogeny and biogeographic history of Lilium–Nomocharis ronmental changes associated with the rapid uplift of Q–T Plateau (ca. 7 Mya) and H-D Event (ca. 3.4 Mya; Chen, 1992). As the orog- The crown clades recovered from the ITS data showed greater eny made available new habitats, distantly related species, such as compatibility with morphology than the matK crown clades. Four those of Nomocharis and Lophophorum, may have developed similar of seven traditional sections delimited on the basis of morphology adaptations to take advantage of them. (Comber, 1949) were recovered in the analyses of ITS. A fifth sec- tion, Sinomartagon (sensu Comber, 1949), was shown to be poly- phyletic, but several small clades contained Sinomartagon 4.2.4. Pseudolirium species exclusively. In contrast to ITS, only two traditional sections The Pseudolirium clade is comprised of North American lilies were recovered in the matK analyses. Additionally, the ITS phylog- that have whorled leaves and turk’s-cap flowers. Our phylogeny eny supports the close relationship of all species of Nomocharis, shows that the clade is sister to Lophophorum, and this is addition- which are highly morphologically distinctive within Lilium (see ally supported by previous karyotype analyses (Gao et al., 2011). Gao et al., 2012). Due to the general concordance between ITS The last shared ancestor of Pseudolirium and Lophophorum oc- and morphology, we have adopted the ITS tree as the best repre- curred within the Q–T plateau region (Fig. 4). The Pseudolirium sentation of the true Lilium–Nomocharis phylogeny. Therefore, the lineage expanded its range into western North America and under- remainder of the discussion within this section refers to the ITS went local extinction in all of eastern Asia (i.e., areas A and B) prior tree unless otherwise noted. to diversification approximately 7 Mya. Within the late Miocene (11.6–5.3 Mya), the Bering land bridge connected East Asia to wes- 4.2.1. Archelirion and Leucolirion tern North America intermittently (Tiffney, 1985; Graham, 1999; This weakly supported crown clade occupies the basal position Wen, 1999) and probably allowed for overland range expansion within Lilium–Nomocharis and is comprised of two robustly sup- of an ancestor of Pseudolirium. ported groups. Archelirion and Luecolirion, which share a trumpet- shaped white colored flower as a synapomorphy, are geographi- 4.2.5. Martagon and Sinomartagon I and II cally isolated from one another and narrowly endemic on the Jap- The eastern Asian Martagon lilies, which also have whorled anese Archipelago and in the H-D Mountains, respectively. The two leaves and turk’s-cap flowers, form monophyletic crown group groups diverged from one another approximately 9.56 (matK, with strong support (Fig. 2). Our ITS phylogeny shows Martagon Fig. 5) 9.87 (ITS, Fig. 4) Mya (12.99–6.02 Mya, 14.24–5.56 Mya, as sister to large clade comprised of Sinomartagon I and II as well HPD, respectively) within the Q–T Plateau region. Archelirion may as the Lilium clade. In the matK gene tree, Sinomartagon I is much have expanded its range onto the Japanese islands shortly after more distantly related to Martagon, which forms a clade within divergence from Leucolirion. Eustatic fluctuations in sea levels dur- Sinomartagon II (Fig. 1). Thus, there is considerable incongruence ing the past 10 Mya (Haq et al., 1987; Kominz et al., 1998) may between gene trees regarding the relationships of species within have permitted migration or dispersal of Archelirion to Japan the traditional sections Martagon and Sinomartagon (sensu Comber, shortly after divergence from Leucolirion. 1949) and between these sections. However, the trees agree on the monophyly of section Martagon and suggest the section has some 4.2.2. Lilium duchartrei and L. lankongense close relationship to Sinomartagon. This may seem surprising since This clade has not been recovered in previous analyses. It is Martagon has been considered significantly more morphologically comprised of two species with similar morphology that have revo- primitive than Sinomartagon (Kim and Lee, 1990) and since com- lute tepals as well as dark purple spots in the flower. These species mercially viable hybrids between these two sections have been lar- were previously recognized as a single taxon (Liang, 1980). How- gely elusive (see Fig. 1 in Van Tuyl et al. (2000)). Despite this, both ever, our analyses of ITS (Fig. 2) support the current classification sections mostly co-occur within eastern Asia (i.e., areas A and B) of Flora of China (Liang and Tamura, 2000), which recognizes them (Liang and Tamura, 2000) and the close relationship between them Y.-D. Gao et al. / Molecular Phylogenetics and Evolution 68 (2013) 443–460 455 has been suggested in previous phylogenetic studies (Nishikawa ular, several stem clades evolved within eastern Asia and the Q–T et al., 1999, 2001; Hayashi and Kawano, 2000). plateau region approximately 7 Mya. Specifically, these were the A common ancestor of Martagon, Sinomartagon I and II and the ancestors of Liriotypus, Nomocharis and other major clades of Lilium Lilium clade probably had a distribution within the H-D Mountains (e.g., Lophophorum, Pseudolirium, ancestral of Matargon and Sino- prior to divergence approximately 8.8 Mya (HPD 6.10–11.81 Mya, martagon, Figs. 4 and 5). Though the eastern margin of the Q–T Pla- Fig. 4). Diversification of major lineages within this group occurred teau may have undergone continuous uplift beginning 20 Mya and around 6–7 Mya and 4–5 Mya. Timing of these events appear con- lasting until approximately 7 Mya, it is widely accepted that sistent with the final phases of Q–T uplift (Harrison et al., 1992) phases of more rapid uplift have occurred (e.g., Harrison et al., and the Hengduan Event (Chen, 1992, 1996). 1992; An et al., 2001; Tapponnier et al., 2001; Song et al., 2001). For example, rapid uplift may have occurred approximately 8– 4.2.6. Lilium 7 Mya and is evident from sedimentary layers indicative of in- In both of our gene phylogenies, there is strong support for the creased aridification (i.e., loess) that accumulated within the rain- Lilium crown clade (Figs. 2 and 3). Species within the clade were shadow of the plateau during that time (Guo et al., 2002). Other first united by Liang (1980) and their close relationship has been lines of geological evidence lead to similar inferences (Harrison supported by previous molecular systematic and karyotype studies et al., 1992). Thus, the timing of the origin of several stem lineages (Nishikawa et al., 1999, 2001; Gao et al., 2012). Members of this within Lilium appears to correspond to the final, rapid phase of Q–T group shared funnel schaped flowers with narrow tubes, as well Plateau uplift (Liu et al., 2006; Zhang et al., 2000; Zhang et al., as white bulbs (Liang, 1980; Liang and Tamura, 2000). 2010; Guo et al., 2002). High rates of biological diversification Based on the ITS divergence time analysis, the Lilium clade arose occurring 8–7 Mya and attributable to Q–T Plateau uplift have also around 4 Mya in East Asia including the Q–T Plateau region (areas A been detected in other eastern Asian organisms from Compositae and B; Fig. 4). In contrast, the clade is younger based on analyses of to catfish (e.g., Liu et al., 2006; Zhang and Fritsch, 2010; Peng matK (2.2 Mya) and evolved in eastern Asia (area B; Fig. 5). In either et al., 2006; Wang et al., 2005). The generation of novel environ- case, vicariance or long distance dispersal may explain the modern, ments along newly-formed elevational gradients may have disjunctive distribution of these lilies between mainland China and prompted ecological speciation in Lilium–Nomocharis. This is con- the Ryukyu-Taiwan island chains. A land route seems plausible be- sistent with the suspected role of hybridization as a driving force cause the Ryukyu-Taiwan Islands and China were connected by dry in the evolution of the complex, especially because hybridization land with a climate suitable for temperate forest (Qian and Ricklefs, is known to occur in the early stages of ecological radiations (Mal- 2000) during the Pleistocene glacial period (ca. 2.6 Mya to 11.7 ka, let, 2007, 2008; Rundle and Nosil, 2005) as well as during periods Gibbard and van Kolfschoten, 2004). Divergence between the insu- of environmental disturbance (Lagache et al., 2013). lar and mainland lineages occurred around 2–1 Mya (ITS and matK Our analyses of divergence time also show that a second phase divergence time dating results, data not shown) and possibility of diversification within Lilium–Nomocharis took place approxi- coincided with the disappearance of a land connection. mately 5–4 Mya, when many eastern Asian and Q–T Plateau crown 4.2.7. Nomocharis and Liriotypus clades evolved (e.g., Martagon, Sinomartagon I and II, Lilium, Figs. 4 Species of the Liriotypus and Nomocharis clades evolved from a and 5). Evolutionary processes such as ecological speciation and common ancestor in the H-D Mountains around 10 Mya (Fig. 4). hybridization, may have occurred again during this time. The tim- Section Liriotypus is a taxonomically stable group (Ikinci_ et al., ing of this second wave of evolutionary events among eastern 2006) with species distributed in the Caucasus and Europe. Our Asian Lilium species roughly corresponds to the H-D Mountain analyses, which included representative Caucasus and European Movement proposed by Chen (1992, 1996). Thus, our data seem species of Liriotypus showed that ancestors of the group occurred to support the conclusions of Chen (1992, 1996); namely that the within Q–T Plateau region well before uplift subsided around H-D Movement may be a distinct, recent phase of Q–T uplift. Dated 7 Mya and were distributed within the paleo-Caucasus region by phylogenies of other Q–T Plateau organisms have also shown spe- at least 7.7 Mya (Fig. 4). In contrast, Nomocharis is a H-D Mountain ciation bursts corresponding to 4 Mya (e.g., Jin et al., 2008; Luo endemic lineage and diversified within the region from a common et al., 2004; Zhang and Fritsch, 2010; Zhang et al., 2009). ancestor that occurred approximately 7 Mya during the late stages Neither the final phase of Q–T uplift nor the H-D orogeny were of Q–T Plateau uplift (Harrison et al., 1992). However, intensive convincingly detected by our analyses of diversification rates diversification of Nomocharis may have occurred within the H-D (Figs. 6 and S1). In the comparatively old Lophophorum and Mountains around 2 Mya (Figs. S1 and S2), possibly following the Nomocharis crown clades (10.22 Mya and 7.32 Mya, respectively; later phase of H-D Mountains orogeny proposed by Chen (1992, Figs. 2 and 4), relatively sharp increases in lineage growth oc- 1996). The relationship between the Liriotypus and Nomocharis curred approximately 6 Mya (Fig. S1) and could be connected to clades (Figs. 2 and 4), was unexpected because of their disjunctive the phase of rapid Q–T uplift from 8–7 Mya. Instead of rapid radi- geographic distributions. ations, we detected high rates of extinction among two of five lin- Discordance between gene trees complicates the position of eages within the Q–T region and declining rates of speciation Nomocharis within the genus Lilium (Figs. 2 and 3). In the matK among two others (Fig. 6). Although somewhat beyond the scope reconstruction, Nomocharis is polyphyletic and both clades are part of the present work, we propose that the high extinction rates of a backbone polytomy (Fig. 3). In the ITS phylogeny, Nomocharis and declining speciation rates may be linked to the evolutionarily is monophyletic if Lilium nepalense, L. saccatum, L. yapingense and L. young ages of the groups in question and their presence in a bio- souliei are included. These Lilium species have none of the synapo- diversity hotspot (i.e., the H-D Mountains; Myers et al., 2000; morphies which have been previously used to delimit Nomocharis Wilson, 1992; http://www.biodiversityhotspots.org/xp/Hotspots; (Sealy, 1950; Gao et al., 2012). However, they share the same habit Richardson et al., 2001a,b). A recent study has shown that loss and distributional range in Q–T Plateau. of biodiversity among recent clades in species rich areas is inde- pendent of anthropogenic effects and may be part of the natural, 4.3. H-D Mountains and its correlation with rapid diversification of stochastic evolutionary course of such species (Davies et al., Lilium in the Q–T Plateau Region 2011). Future studies using Lilium–Nomcharis as a model system may seek to better understand the relationships among geomor- Based on our analyses of divergence times, two distinct phases phology, evolutionary rates, and extinctions within the Q–T Pla- of diversification can be detected in eastern Asian Lilium. In partic- teau and H-D Mountains regions. 456 Y.-D. Gao et al. / Molecular Phylogenetics and Evolution 68 (2013) 443–460

5. Conclusions ITS accessions: Cardiocrinum cathayanum (E.H. Wilson) Stearn, HM045474; Cardiocrinum giganteum (Wallich) Makino, Our results confirm previous reports that traditional sections of HM045473; Fritillaria cirrhosa D. Don, HM045469; Fritillaria thun- Lilium are paraphyletic and that Nomocharis is nested within a bergii Miquel, HQ448863; Fritillaria unibracteata P.K. Hsiao et K.C. broader Lilium–Nomocharis complex. Despite several points of Hsia, HQ448866; Lilium akkusianum Gämperle, AM292422; Lilium agreement, significant discordance was detected between ITS and albanicum Griseb., AM292432; Lilium albanicum Griseb., matK gene trees. The ITS phylogeny could generally be supported EF042793; Lilium alexandrae Couts, AB020475; Lilium amabile Pal- by morphology, while the matK phylogeny showed geographically ibin, HQ456828; Lilium amoenum Wilson, AB035284; Lilium anhu- partitioned clades. Based on the available data, we conclude that iense D.C. Zhang et J.Z. Shao, HM045454; Lilium armenum (Miscz. hybridization, rather than incomplete lineage sorting has been an ex Grossh.) Manden., AM292425; Lilium artvinense Miscz., important evolutionary mechanism in Lilium and may be responsi- AM292427; Lilium auratum Lindley, AB020472; Lilium auratum ble for discordance between the chloroplast and nuclear trees. Lindley var. platyphyllum Baker, AB020474; Lilium bakerianum Coll- However, continued work on the purported hybrid, N. gongshanen- ett & Hemsley, HM045428; Lilium bakerianum Collett & Hemsley sis, as well as on reconciling gene and species trees within the var. rubrum Stearn, HQ456829; Lilium bakerianum Collett et Hems- whole group, is certain to provide additional insights. Biogeo- ley var. delavayi (Franchet) E.H. Wilson, HM045468; Lilium bolan- graphic analyses concur with previous studies at higher taxonomic deri S. Watson, AB035278; Lilium bosniacum Beck ex Fritsch, levels, which have suggested that Lilium evolved within the Q–T AM292423; Lilium bosniacum Beck ex Fritsch, EF042788; Lilium Plateau region. Finally, our results support two waves of diversifi- brownii F. E. Brown ex Miellez var. viridulum Baker, HQ692117; Lili- cation of eastern Asian Lilium. One of these corresponds to a period um bulbiferum L., AB020468; Lilium callosum Siebold et Zuccarini, of rapid, late Q–T Plateau uplift around 8–7 Mya. A second wave, AB020471; Lilium canadense L., AB020457; Lilium candidum L., occurring around 4 Mya, appears temporally correlated with a dis- EF042778; Lilium carneolicum Bernardi & Koch, AM292419; Lilium tinct H-D Movement. carneolicum Bernardi & Koch, EF042782; Lilium cernuum Komarov, HM045427; Lilium chalcedonicum L., EF042781; Lilium ciliatum P.H. Davis, AM292421; Lilium columbianum Hanson ex Baker, Acknowledgments AF090963; Lilium concolor Salisbury var. pulchellum (Fischer) Regel, HM045460; Lilium dauricum Ker Gawler, HM045446; Lilium davidii The authors thank Dr. You-Sheng Chen and Xiao-Hua Jin from Duchartre ex Elwes, HQ692078; Lilium distichum Nakai ex Kami- Institution of Botany, Chinese Academy of Sciences for the help bayashi, HM045451; Lilium duchartrei Franchet, HQ692064; Lilium in field work and material collection. This research was supported duchartrei Franchet, HQ448864; Lilium fargesii Franchet, by the National Natural Science Foundation of China (Grant Nos. HM045459; Lilium formosanum Wallace, AB020470; Lilium grayi 31100161, 31270241 and 31070166), and the specimen platform Watson, AF090961; Lilium hansonii D.T. Moore, AB020465; Lilium of China, teaching specimen’s sub-platform (http://mnh.scu.e- henrici Franchet, HM045456; Lilium henryi Baker, HM045462; Lili- du.cn/). A National Science Foundation (NSF) East Asia and Pacific um humboldtii Duchartre, AY616746; Lilium jankae A. Kern, Summer Institutes (EAPSI) Fellowship awarded to Harris (Award EF042791; Lilium japonicum Thunberg ex Houttuyn, AB020451; ID: OISE-1209651) has helped to facilitate international collabora- Lilium jinfushanense L.J. Peng et B.N. Wang, HQ692157; Lilium kes- tion associated with this research. selringianum Mischenko, AM292429; Lilium lankongense Franchet, HQ692145; Lilium lankongense Franchet, HM045430; Lilium leich- Appendix A. tlinii J.D. Hooker var. maximowiczii (Regel) Baker, AB020454; Lilium leucanthum (Baker) Baker, HM045466; Lilium leucanthum (Baker) The sequences used in the present study are presented by mar- Baker var. centifolium (Stapf ex Elwes) Stearn, HM045463; Lilium ker and are ordered as species, Genbank accession number. Acces- lijiangense L.J. Peng, HM045424; Lilium longiflorum Thunberg, sions new to this study are presented in bold type. AB020458; Lilium longiflorum Thunberg var. scabrum Masamune, rbcL accessions: Alstroemeria aurea Graham, AY465703; HM045447; Lilium lophophorum (Bureau & Franchet) Franchet, Androcymbium ciliolatum Schltr. & Krause, Z77265; Bomarea hir- HQ692099; Lilium mackliniae Sealy, AB035286; Lilium maculatum tella Herb., Z77255; Burchardia umbellate R. Br., Z77266; Calo- Thunberg var. monticola Hara, AB020460; Lilium maritimum Kel- chortus balsensis García-Mend., AF275985; Calochortus weedii logg, AY616748; Lilium martagon L., AB020455; Lilium martagon Alph Wood, AF275987; Campynema lineare Labill., Z77264; L., EF042777; Lilium martagon Linnaeus var. pilosiusculum Freyn, Campynemanthe viridiflora Baill., AJ276349; Cardiocrinum gigante- HM045452; Lilium matangense J.M. Xu, HM045457; Lilium medeol- um Makino, AF275988; Clintonia borealis Raf., AB056856; Clinto- oides A. Gray, AB020448; Lilium michiganense Farwell, AB020440; nia umbellulata (Michx.) Morong, AB056854; Drymophila moorei Lilium monadelphum Bieberstein, AM292418; Lilium nanum Baker, AB088812; Erythronium japonicum Poit., D28156; Fritil- Klotzsch in Klotzsch & Garcke, HQ687289; Lilium nanum Klotzsch laria meleagris L., AY395537; Gagea wilczekii Braun-Blanq. & in Klotzsch et Garcke var. flavidum (Rendle) Sealy, HM045458; Lili- Maire, AF275990; Lapageria rosea Ruiz & Pav., Z77301; Lilium um nepalense D. Don, HQ687293; Lilium nevadense Eastw., speciosum Thunb., AB034922; Lloydia serotina (L.) Salisb. ex AB035279; Lilium nobilissimum T. Makino, AB020450; Lilium occi- Rchb., Z77294; Luzuriaga radicans Ruiz & Pav., Z77300; Medeola dentale Purdy, AF090965; Lilium ocellatum (Kellogg) Beane, virginiana L., AY465706; Medeola virginiana L., M91631; Nomo- AF090956; Lilium oxypetalum Baker, AB020442; Lilium papilliferum charis pardanthina Franch., Z77295; Notholirion bulbuliferum (Lin- Franchet, HQ687262; Lilium pardalinum Kellog, AB020439; Lilium gelsh.) Stearn, AF275991; Philesia magellanica J.F. Gmel., Z77302; pardalinum Kellog, AB020452; Lilium pardalinum Kellogg var. Rhipogonum elseyanum F. Muell., Z77309; Smilax aspera DC., giganteum Stearn & Woodcock, AB020445; Lilium parryi S. Watson, GU945049; Smilax china L., D28333; Smilax glauca Mart., AB020435; Lilium parvum Kellogg, AB020436; Lilium philippinense AF206822; Smilax rotundifolia L., AY465710; Tricyrtis affinis Mak- Baker, AB020437; Lilium pomponium L., EF042779; Lilium ponticum ino, D17382; Tricyrtis latifolia Maxim., AF275993; Trillium chlo- Koch, AM292426; Lilium primulinum Baker var. burmanicum (W.W. ropetalum (Torr.) Howell, AB018837; Tripladenia cunninghamii Smith) Stearn, HM045449; Lilium primulinum Baker var. ochraceum D. Don, Z77268; Tulipa kolpakowskiana Regel, Z77292; Uvularia (Franchet) Stearn, HM045450; Lilium pumilum Redouté, puberula Michx., AB009952; Xerophyllum tenax (Pursh) Nutt., HQ692084; Lilium pyrenaicum Gouan, EF042780; Lilium regale AJ131949. E.H. Wilson, HQ692090; Lilium rubellum Baker, AB020429; Lilium Y.-D. Gao et al. / Molecular Phylogenetics and Evolution 68 (2013) 443–460 457 rubescens Watson, AY616749; Lilium saccatum S.Y. Liang, JQ724596; Lilium matangense J.M. Xu, JQ724584; Lilium medeolo- HQ687291; Lilium saccatum S.Y. Liang, HQ687292; Lilium sargenti- ides A. Gray, AB030873; Lilium medeoloides A. Gray, AB049495; ae E.H. Wilson, HQ692112; Lilium sempervivoideum H. Léveillé, Lilium michiganense Farwell, AB030844; Lilium nanum Klotzsch HM045467; Lilium souliei (Franchet) Sealy, JQ724631; Lilium spec- in Klotzsch et Garcke, JQ724598; Lilium nepalense D. Don, iosum Thunberg var. clivorum S. Abe et T. Tamura, AB020431; Lili- JQ724585; Lilium nevadense Eastw., AB049516; Lilium nobilissi- um speciosum Thunberg var. gloriosoides Baker, HM045461; Lilium mum T. Makino, AB030851; Lilium nobilissimum T. Makino, sulphureum Baker ex J.D. Hooker, HQ692124; Lilium superbum L., AB049502; Lilium ocellatum Elwes, AB049512; Lilium oxypetalum AB020420; Lilium szovitsianum Fischer & Ave-Lallemant, Baker, AB049498; Lilium pardalinum Kellog, AB030845; Lilium par- AM292428; Lilium taliense Franchet, HQ692109; Lilium tigrinum dalinum Kellog, AB049508; Lilium pomponium L., AB030865; Lili- Ker Gawle, HQ692093; Lilium tsingtauense Gilg., HQ687259; Lilium um primulinum Baker var. burmanicum (W.W. Smith) Stearn, wallichianum Schultes et Schultes, AB020422; Lilium wardii Stapf JQ724611; Lilium primulinum Baker var. ochraceum (Franchet) ex F.C. Stern, AB035287; Lilium washingtonianum Kellog, Stearn, JQ724581; Lilium pumilum Redouté, JQ724592; Lilium AB020438; Lilium wenshanense L.J. Peng et F.X. Li, HM045453; Lili- pyrenaicum Gouan, AB030866; Lilium pyrenaicum Gouan, um xanthellum F.T. Wang et Tang, HQ692154; Lilium xanthellum F.T. AB049494; Lilium regale E.H. Wilson, JQ724624; Lilium rosthornii Wang et Tang var. luteum S.Y. Liang, HQ692152; Lilium yapingense Diels, AB030861; Lilium rubellum Baker, AB030852; Lilium rubel- Y.D. Gao et X.J. He, HQ687290; Nomocharis aperta (Franchet) E.H. lum Baker, AB049521; Lilium saccatum S.Y. Liang, JQ724586; Lili- Wilson, JQ724632; Nomocharis aperta (Franchet) E.H. Wilson, um sargentiae E.H. Wilson, JQ724576; Lilium sempervivoideum H. HM045433; Nomocharis basilissa Farrer ex W.E. Evans, Léveillé, JQ724612; Lilium speciosum Thunberg, AB030853; Lilium HQ687260; Nomocharis farreri (W.E. Evans) Harrow, HM045437; speciosum Thunberg var. clivorum S. Abe et T. Tamura, AB049504; Nomocharis gongshanensis Y.D. Gao et X.J. He, HM045438; Nomo- Lilium speciosum Thunberg var. gloriosoides Baker, JQ724613; Lili- charis gongshanensis Y.D. Gao et X.J. He, HM045442; Nomocharis um sulphureum Baker ex J.D. Hooker, JQ724614; Lilium superbum meleagrina Franchet, HM045436; Nomocharis pardanthina Fran- L., AB024546; Lilium taliense Franchet, JQ724590; Lilium tigrinum chet, HM045432; Nomocharis pardanthina Franchet, JQ724635; Ker Gawle, JQ724577; Lilium tsingtauense Gilg., JQ724599; Lilium Nomocharis pardanthina Franchet f. punctulata Sealy, HM045435; washingtonianum Kellog, AB030848; Lilium wenshanense L.J. Peng Nomocharis pardanthina Franchet f. punctulata Sealy, JQ724633; et F.X. Li, JQ724603; Lilium xanthellum F.T. Wang et Tang, Nomocharis pardanthina Franchet, HM045431; Nomocharis saluen- JQ724588; Lilium xanthellum F. T. Wang et Tang var. luteum S. ensis I.B. Balfour, HM045434; Nomocharis sp., JQ724634; Notholiri- Yun Liang, JQ724589; Lilium yapingense Y.D. Gao et X.J. He, on bulbuliferum (Lingelsheim ex H. Limpricht) Stearn, HQ448856; JQ724587; Nomocharis aperta (Franchet) E.H. Wilson, JQ724616; Notholirion macrophyllum (D. Don) Boissier, HM045475. Nomocharis basilissa Farrer ex W.E. Evans, JQ724621; Nomocharis matK accessions: Cardiocrinum cathayanum (E.H. Wilson) farreri (W.E. Evans) Harrow, JQ724618; Nomocharis gongshanensis Stearn, JQ724628; Cardiocrinum giganteum (Wallich) Makino, Y.D. Gao et X.J. He, JQ724619; Nomocharis meleagrina Franchet, JQ724627; Fritillaria cirrhosa D. Don, JQ724625; Fritillaria thun- JQ724617; Nomocharis pardanthina Franchet, AB030842; Nomo- bergii Miquel, JQ724626; Lilium alexandrae Couts, AB030849; Lili- charis pardanthina Franchet, JQ724622; Nomocharis pardanthina um alexandrae Couts, AB049490; Lilium amabile Palibin, Franchet f. punctulata Sealy, JQ724615; Nomocharis saluenensis JQ724593; Lilium amoenum Wilson, AB049515; Lilium anhuiense I.B. Balfour, JQ724623; Nomocharis sp., JQ724620; Notholirion D.C. Zhang et J.Z. Shao, JQ724600; Lilium auratum Lindley, bulbuliferum (Lingelsheim ex H. Limpricht) Stearn, JQ724630; AB049522; Lilium auratum Lindley var. platyphyllum Baker, Notholirion macrophyllum (D. Don) Boissier, JQ724629. AB049520; Lilium bakerianum Collett & Hemsley, JQ724602; Lili- um bakerianum Collett & Hemsley var. rubrum Stearn, Appendix A. Supplementary material JQ724601; Lilium brownii F.E. Brown ex Miellez var. viridulum Ba- ker, JQ724604; Lilium bulbiferum L., AB030864; Lilium callosum Supplementary data associated with this article can be found, in Siebold et Zuccarini, AB030854; Lilium canadense L., AB030843; the online version, at http://dx.doi.org/10.1016/j.ympev.2013.04. Lilium candidum L., AB024545; Lilium cernuum Komarov, 026. JQ724595; Lilium ciliatum P.H. Davis, AB049500; Lilium columbia- num Hanson ex Baker, AB030847; Lilium concolor Salisbury var. References pulchellum (Fischer) Regel, JQ724605; Lilium dauricum Ker Gawler, AB049497; Lilium davidii Duchartre ex Elwes, JQ724606; Lilium Allen, G.A., 2001. Hybrid speciation in Erythronium (Liliaceae), a new allotetraploid distichum Nakai ex Kamibayashi, JQ724594; Lilium duchartrei species from Washington State. Systematic Botany 26, 263–272. An, Z.S., Kutzbach, J.E., Prell, W.L., Porter, S.C., 2001. Evolution of Asian monsoons Franchet, JQ724578; Lilium fargesii Franchet, JQ724582; Lilium for- and phased uplift of the Himalaya-Tibetan plateau since Late Miocene times. mosanum Wallace, AB030867; Lilium hansonii D.T. Moore, Nature 411, 62–66. AB049519; Lilium henrici Franchet, JQ724597; Lilium henryi Baker, Antonelli, A., Nylander, J.A.A., Persson, C., Sanmartín, I., 2009. Tracing the impact of the Andean uplift on Neotropical plant evolution. Proceedings of the National JQ724583; Lilium humboldtii Duchartre, AY624461; Lilium japoni- Academy of Sciences 106, 9749–9754. cum Thunberg ex Houttuyn, AB030850; Lilium japonicum Thun- Baker, J.G., 1871. A new synopsis of all the known lilies. Gardeners Chronicle 104, berg ex Houttuyn, AB049514; Lilium japonicum Thunberg ex 1650. Balfour, B., 1918. The genus Nomocharis. Botanical Journal of Scotland 27 (3), 273– Houttuyn var. abeanum Kitamura, AB049528; Lilium jinfushanense 300. L.J. Peng et B.N. Wang, JQ724591; Lilium lankongense Franchet, Baranova, M.V., 1969. The geographical distribution of Lilium species in the flora of JQ724607; Lilium leichtlinii Hooker f. var. maximowiczii Baker, the USSR. Lily Year Book RHS 32, 39–55. AB030860; Lilium leucanthum (Baker) Baker var. centifolium (Stapf Baranova, M.V., 1988. A synopsis of the system of the genus Lilium (Liliaceae). Botanicheskii Zhurnal 73, 1319–1329. ex Elwes) Stearn, JQ724579; Lilium leucanthum (Baker) Baker, Berry, E.W., 1930. Revision of the Lower Eocene Wilcox Flora of the Southeastern JQ724580; Lilium lijiangense L.J. Peng, JQ724608; Lilium longiflo- States: with Descriptions of New Species, chiefly from Tennessee and Kentucky, rum Thunberg var. scabrum Masamune, JQ724609; Lilium lopho- vol. 156. Geological Survey Professional Papers. U.S., pp. 1–196 Bews, J.W., 1927. Studies in the ecological evolution of the angiosperms. New phorum (Bureau & Franchet) Franchet, JQ724610; Lilium Phytologist 26, 1–21. mackliniae Sealy, AB030877; Lilium maculatum Thunberg, Bremer, K., 2000. Early Cretaceous lineages of monocot flowering plants. AB030875; Lilium maculatum Thunberg var. monticola Hara, Proceedings of the National Academy of Sciences United States of America 97 (9), 4707–4711. AB049518; Lilium maritimum Kellogg, AY624463; Lilium martagon Buckler, E.S., Holtsford, T.P., 1996. Zea systematics: ribosomal ITS evidence. L., AB030872; Lilium martagon Linnaeus var. pilosiusculum Freyn, Molecular Biology and Evolution 13 (4), 612–622. 458 Y.-D. Gao et al. / Molecular Phylogenetics and Evolution 68 (2013) 443–460

Carpenter, R.J., Jordan, G.J., Hill, R.S., 2007. A toothed Lauraceae leaf from the early Harpke, D., Peterson, A., 2008. 5.8S motifs for the identification of pseudogenic ITS Eocene of Tasmania, Australia. International Journal of Plant Sciences 168, regions. Botany 86, 300–305. 1191–1198. Harrison, T.M., Copeland, P., Kidd, W.S.F., Yin, A., 1992. Raising Tibet. Science 255, Che, J., Zhou, W.-W., Hu, J.-S., Yan, F., Papenfuss, T.J., Wake, D.B., Zhang, Y.-P., 2010. 1663–1670. Spiny frogs (Paini) illuminate the history of the Himalayan region and Southeast Haw, S.G., 1986. The Lilies of China. Timber Press, Portland. Asia. Proceedings of the National Academy of Sciences 107, 13765–13770. Hayashi, K., Kawano, S., 2000. Molecular systematics of Lilium and allied genera Chen, F.B., 1992. H-D event: an important tectonic event of the late Cenozoic in (Liliaceae): phylogenetic relationships among Lilium and related genera based Eastern Asia. Mountain Research 10 (4), 195–202. on the rbcL and matK gene sequence data. Plant Species Biology 15, 73–93. Chen, F.B., 1996. Second discussion on the H-D movement. Mountain Research 17 Hills, L., Baadsgaard, H., 1967. Potassium-argon dating of some lower Tertiary strata (3–4), 14–22. in British Columbia. Bulletin of Canadian Petroleum Geology 15, 138–149. Chung, S.L., Lo, C.H., Lee, T.Y., Zhang, Y.Q., Xie, Y.W., Li, X.H., Wang, K.L., Wang, P.L., Ho, S.Y.W., Phillips, M.J., 2009. Accounting for calibration uncertainty in 1998. Diachronous uplift of the Tibetan plateau starting 40 Myr ago. Nature phylogenetic estimation of evolutionary divergence times. Systematic Biology 394, 769–773. 58 (3), 367–380. Cockerell, T.D.A., 1922. A new genus of fossil Liliaceae. Bulletin of the Torrey Homes, W., 1993+. Smilacaceae. In: Flora of North America Editorial Committee Botanical Club 49, 211–213. (Ed.), Flora of North America: Magnoliophyta: Liliidae: Liliales and Orchidales. Comber, H.F., 1949. A new classification of genus Lilium. Royal Horticultural Society Oxford University Press, Oxford. of the Liliy Year Book 13, 85. Ikinci,_ N., 2011. Molecular phylogeny and divergence times estimates of Lilium Comes, H.P., Abbott, R.J., 2001. Molecular phylogeography, reticulation, and lineage section Liriotypus (Liliaceae) based on plastid and nuclear ribosomal ITS DNA sorting in Mediterranean Senecio sect. Senecio (Asteraceae). Evolution 55, 1943– sequence data. Turkish Journal of Botany 35, 319–330. 1962. Ikinci,_ N., Oberprieler, C., Güner, A., 2006. On the origin of European lilies: Conran, J.G., Carpenter, R.J., Jordan, G.J., 2009. Early Eocene Ripogonum (Liliales: phylogenetic analysis of Lilium section Liriotypus (Liliaceae) using sequences of Ripogonaceae) leaf macrofossils from southern Australia. Australian Systematic the nuclear ribosomal transcribed spacers. Willdenowia 36, 647–656. Botany 22, 219–228. Jin, Y., Brown, R.P., Liu, N., 2008. Cladogenesis and phylogeography of the lizard Cuénoud, P., Savolainen, V., Chatrou, L.W., Powell, M., Grayer, R.J., Chase, M.W., Phrynocephalus vlangalii (Agamidae) on the Tibetan plateau. Molecular Ecology 2002. Molecular phylogenetics of Caryophyllales based on nuclear 18S rDNA 17, 1971–1982. and plastid rbcL, atpB, and matK DNA sequences. American Journal of Botany Joly, S., McLenachan, P.A., Lockhart, P.J., 2009. A statistical approach for 89, 132–144. distinguishing hybridization and incomplete lineage sorting. The American Daghlian, C.P., 1981. A review of the fossil record of monocotyledons. Botanical Naturalist 174 (2), E54–E70. Review 47 (4), 517–555. Kelchner, S.A., 2000. The evolution of non-coding chloroplast DNA and its Davies, T.J., Smith, G.F., Bellstedt, D.U., Boatwright, J.S., Bytebier, B., Cowling, R.M., application in plants systematics. Annals of the Missouri Botanical Garden 87, Forest, F., Harmon, L.J., Muasya, A.M., Schrire, B.D., Steenkamp, Y., van der Bank, 482–498. M., Savolainen, V., 2011. Extinction risk and diversification are linked in a plant Kim, Y.S., Lee, W.B., 1990. A study of the morphological characters of the genus biodiversity hotspot. PLoS Biology 9 (5), e1000620. Lilium L. Korea. Korean Journal of Plant Taxonomy 20 (3), 165–178. Dilcher, D.L., Lott, T.A., 2005. A middle Eocene fossil plant assemblage (Powers Clay Kominz, M.A., Miller, K.G., Browning, J.V., 1998. Long-term and short-term global Pit) from western Tennessee. Bulletin Florida Museum Natural History 45, 1– Cenozoic sea-level estimates. Geology 26, 311–314. 43. Lagache, L., Klein, E.K., Guichoux, E., Petit, R.J., 2013. Fine-scale environmental Ding, S.T., Sun, B.N., Wu, J.Y., Li, X.C., 2011. Miocene Smilax leaves and associated control of hybridization in oaks. Molecular Ecology 22 (2), 423–424. epiphyllous fungi from Zhejiang, East China and their paleoecological Lee, C.S., Kim, S.C., Yeau, S.H., Lee, N.S., 2011. Major lineages of the genus Lilium implications. Review of Palaeobotany and Palynology 165, 209–223. (Liliaceae) based on nrDNA ITS sequences, with special emphasis on the Korean Douglas, N.A., Wall, W.A., Xiang, Q.Y., Hoffman, W.A., Wentworth, T.R., Gray, J.B., species. Journal of Plant Biology 54, 159–171. Hohmann, M.G., 2011. Recent vicariance and the origin of the rare, edaphically Li, H., 1980. Himalayas-H-D Mountains – the center of distribution and specialized Sandhills lily, Lilium pyrophilum (Liliaceae): evidence from differentiation of the genus Arisaema – to discuss the problems about the phylogenetic and coalescent analyses. Molecular Ecology 20, 2901–2915. origin and migration of this genus. Acta Botanica Yunnanica 2 (4), 402–416. Doyle, J.A., 1973. Fossil evidence on early evolution of the monocotyledons. The Li, J.J., Shi, Y.F., Li, B.Y., 1995. Uplift of the Qinghai-Xizang (Tibet) Plateau and Global Quarterly Review of Biology 48 (3), 399–413. Change. Lanzhou Univ Press, Lanzhou Doyle, J.J., Doyle, J.L., 1987. A rapid DNA isolation procedure for small quantities of Liang, S.Y., 1980. Lilium L. In: Wang F.T., Tang T. (eds.), Flora Reipublicae Popularis fresh leaf tissue. Phytochemical Bulletin 19, 11. Sinicae, vol. 14. Anagiospermae, Monocotyledoneae Liliaceae (I). Science Press, Drummond, A.J., Rambaut, A., 2007. Beast: Bayesian evolutionary analysis by Beijing, pp. 116-157. sampling trees. BMC Evolutionary Biology 7, 214. Liang, S.Y., 1984. Studies on the genus Nomocharis (Liliaceae). Bulletin of Botanical Drummond, A.J., Ho, S.Y.W., Phillips, M.J., Rambaut, A., 2006. Relaxed phylogenetics Research 4 (3), 163–178. and dating with confidence. PLoS Biology 4 (5), e88. Liang, S.Y., 1995. Chorology of Liliaceae (S. Str.) and its bearing on the Chinese Flora. Edgar, R.C., 2004. MUSCLE: multiple sequence alignment with high accuracy and Acta Phytotaxonomica Sinica 33 (1), 27–51. high throughput. Nucleic Acids Research 32 (5), 1792–1797. Liang, S.Y., Tamura, M., 2000. In: Wu, Z.Y., Raven, P.H. (Eds.), Flora of China, vol. 24. Erwin, D.M., Stockey, R.A., 1991. Soleredera rhizomorpha gen. et sp. nov., a Science Press Beijing, Missouri Botanical Garden Press, St. Louis, pp. 135–159. permineralized monocotyledon from the middle Eocene Princeton Chert of Lighty, R.W., 1960. Cytological and interspecific hybridization studies in Lilium L. British Columbia, Canada. Botanical Gazette 152, 231–247. and their significance for classification. PhD thesis, Cornell University, New Evanoff, E., Mcintosh, W., Murphey, P., 2001. Stratigraphic summary and ‘ArP’ Ar York. geochronology of the Florissant Formation, Colorado. In: Evanoff, E., Gregory- Lighty, R.W., 1968. Evolutionary trends in lilies. Lily Year Book RHS 31, 40–44. Wodzicki, K., Johnson, K. (Eds.), Fossil Flora and Stratigraphy of the Florissant Linder, C.R., Rieseberg, L.H., 2004. Reconstructing patterns of reticulate evolution in Formation, Colorado. plants. American Journal of Botany 91, 1700–1708. Evans, W.E., 1925. A revision of the genus Nomocharis. Notes form the Royal Botanic Liu, J.-Q., Wang, Y.-J., Wang, A.-L., Hideaki, O., Abbott, R.J., 2006. Radiation and Garden Edinburgh 15, 1–46. diversifcation within the Ligularia–Cremanthodium–Parasenecio complex Farris, J.S., Källersjö, M., Kluge, A.G., Bult, C., 1994. Testing significance of (Asteraceae) triggered by uplift of the Qinghai–Tibetan Plateau. Molecular incongruence. Cladistics 10, 315–319. Phylogenetics and Evolution 38, 31–49. Fehrer, J., Krahulcová, A., Krahulec, F., Chrtek Jr., J., Rosenbaumová, R., Bräutigam, S., Luo, J., Yang, D., Suzuki, H., Wang, Y., Chen, W.-J., Campbell, K.L., Zhang, Y.-P., 2004. 2007. Evolutionary aspects in Hieracium subgenus Pilosella. In: Grossniklaus, U., Molecular phylogeny and biogeography of Oriental voles: genus Eothenomys Hörandl, E., Sharbel, T., van Dijk, P. (Eds.), Apomixis: Evolution, Mechanisms and (Muridae, Mammalia). Molecular Phylogenetics and Evolution 33, 349– Perspectives (Regnum Vegetabile). Koeltz, Königstein, pp. 359–390. 362. Franchet, A.R., 1889. Nomocharis Franchet. Journal de Botanique 3, 113. Maddison, W., Knowles, L.L., 2006. Inferring phylogeny despite incomplete lineage Friis, E., Crane, P.R., Pedersen, K.R., 2011. Early Flowers and Angiosperm Evolution. sorting. Systematic Biology 55 (1), 21–30. Cambridge University Press, Cambridge. Maddison, W.P., Maddison, D.R., 2011. Mesquite: A Modular System for Gao, Y.D., Zhou, S.D., He, X.J., 2011. Karyotype studies of thirty-two species in the Evolutionary Analysis. 2.75. . genus Lilium (Liliaceae) of China. Nordic Journal of Botany 29, 746–761. Mallet, J., 2007. Hybrid speciation. Nature 446, 279–283. Gao, Y.D., Hohenegger, M., Harris, A., Zhou, S.D., He, X.J., Wan, J., 2012. A new species Mallet, J., 2008. Hybridization, ecological races and the nature of species: empirical in the genus Nomocharis Franchet (Liliaceae): evidence that brings the genus evidence for the ease of speciation. Philosophical Transactions of the Royal Nomocharis into Lilium. Plant Systematics and Evolution 298, 69–85. Society B: Biological Sciences 363, 2971–2986. Gibbard, P., van Kolfschoten, T., 2004. The Pleistocene and Holocene Epochs. In: Manchester, S.R., 2001. Update on the megafossil flora of Florissant, Colorado, USA. Gradstein, F.M., James, G.O., Gilbert, S.A. (Eds.), A Geologic Time Scale 2004. In: Evanoff, E., Gregory-Wodzicki, K.M., Johnson, K.R. (Eds.), Fossil Flora and Cambridge University Press, Cambridge, UK. Stratigraphy of the Florissant Formation. Colorado. Graham, A., 1999. Late Cretaceous and Cenozoic history of North American Mason-Gamer, R.J., Holsinger, K.E., Jansen, R.K., 1995. Chloroplast DNA haplotype Vegetation North of Mexico. Oxford University Press, New York, USA. variation within and among populations of Coreopsis grandiflora (Asteraceae). Guo, Z.T., Ruddiman, W.F., Hao, Q.Z., Wu, H.B., Qiao, Y.S., Zhu, R.X., Peng, S.Z., Wei, Molecular Biology and Evolution 12, 371–381. J.J., Yuan, B.Y., Liu, T.S., 2002. Onset of Asian desertification by 22 Myr ago McRae, E.A., 1998. Lilies: A Guide for Growers and Collectors. Timber Press, inferred from loess deposits in China. Nature 416, 159–163. Portland, pp. 17–26. Haq, B.U., Hardenbol, J., Vail, P.R., 1987. Chronology of fluctuating sea levels since Moens, P.B., 1969. The fine structure of meiotic chromosome pairing in the triploid, the Triassic. Science 235, 1156–1167. Lilium tigrinum. The Journal of Cell Biology 40, 273–279. Y.-D. Gao et al. / Molecular Phylogenetics and Evolution 68 (2013) 443–460 459

Molvary, M.P., Kores, P.J., Chase, M.W., 2000. Polyphyly of mycoheterotrophic Shi, Y.F., Li, J.J., Li, B.Y., 1998. Uplift and Environmental Changes of Qinghai–Tibetan orchids and functional influences on floral and molecular characters. In: Wilson, Plateau in the Late Cenozoic. Guangdong Science and Technology Press, K.L., Morrison, D.A. (Eds.), Monocots: Systematics and Evolution. CSIRO Guangzhou. Publishing, Collingwood, Victoria, Australia, pp. 441–448. Shinwari, Z.K., Shinwari, S., 2010. Molecular data and phylogeny of family Moore, W.S., 1995. Inferring phylogenies from mtDNA variation: Mitochondrial- Smilacaceae. Pakistan Journal of Botany 42, 111–116. gene trees versus nuclear-gene trees. Evolution 49, 718–726. Skinner, M.W., 1988. Comparative Pollination Ecology and Floral Evolution in Pacific Muller, J., 1981. Fossil pollen records of extant angiosperms. Botanical Review 47, Coast Lilium. Ph.D. Dissertation. Harvard University. 1–142. Song, C., Fang, X., Li, J., Gao, J., Zhao, Z., Fan, M., 2001. Tectonic uplift and Muratovic´, E., Hidalgo, O., Garnatje, T., Siljak-Yakovlev, S., 2010. Molecular sedimentary evolution of the Jiuxi Basin in the northern margin of the Tibetan phylogeny and genome size in European Lilies (Genus Lilium, Liliaceae). Plateau since 13 Ma BP. Science in China Series D: Earth Sciences 44, 192– Advanced Science Letters 3 (2), 180–189. 202. Myers, N., Mittermeier, R.A., Mittermeier, C.G., da Fonseca, G.A.B., Kent, J., 2000. Spicer, R.A., Harris, N.B., Widdowson, W.M., Herman, A.B., Guo, S., Valdes, P.J., Wolfe, Biodiversity hotspots for conservation priorities. Nature 403, 853–858. J.A., Kelley, S.P., 2003. Constant elevation of southern Tibet over the past Nishikawa, T., Okazaki, K., Uchino, T., Arakawa, K., Nagamine, T., 1999. A molecular 15 million years. Nature 421, 622–624. phylogeny of Lilium in the internal transcribed spacer region. Journal of Stevens, P.F., 2001 onwards. Angiosperm Phylogeny Website Version 9. (accessed 02.08.10). Nishikawa, T., Okazaki, K., Arakawa, K., Nagamine, T., 2001. Phylogenetic analysis of Stewart, R.N., Bamford, R., 1943. The nature of polyploidy in Lilium tigrinum. section Sinomartagon in genus Lilium using sequences of the internal American Journal of Botany 30, 1–7. transcribed spacer region in nuclear ribosomal DNA. Breeding Science 51 (1), Stigall Rode, A.L., Lieberman, B.S., 2005. Paleobiogeographic patterns in the Middle 39–46. and Late Devonian emphasizing Laurentia. Palaeogeography, Nylander, J.A.A., 2004. MrModeltest 2.0. Program distributed by the author. Palaeoclimatology, Palaeoecology 222, 272–284. Department of Systematic Zoology, EBC, Uppsala University, Uppsala. Stockey, R.A., Wehr, W.C., 1996. Flowering plants in and around the Eocene lakes of Pagel, M., Meade, A., Barker, D., 2004. Bayesian estimation of ancestral character the interior. In: Ludvigsen, R. (Ed.), Life in Stone: A Natural History of British states on phylogenies. Systematic Biology 53, 673–684. Columbia’ Fossils. UBC Press, Vancouver, BC. Pan, Y.S., 1989. Division of geologic structure in the H-D Mountains region. Journal Swofford, D.L., 2003. PAUP: Phylogenetic Analysis Using Parsimony (and Other of Mountain Research 7 (1), 3–12. Methods). Version 4. Sinauer Associates, Sunderland, Massachusetts. Pan, G., Wang, L., Li, R., Yuan, S., Ji, W., Yin, F., Zhang, W., Wang, B., 2012. Tectonic Takhtajan, A., 1997. Diversity and Classification of Flowering Plants. Columbia evolution of the Qinghai–Tibet Plateau. Journal of Asian Earth Sciences. University Press, New York. Patterson, T.B., Givnish, T.J., 2002. Phylogeny, concerted convergence, and Tamura, K., Dudley, J., Nei, M., Kumar, S., 2007. MEGA4: Molecular Evolutionary phylogenetic niche conservatism in the core Liliales: insights from rbcL and Genetics Analysis (MEGA) software version 4.0. Molecular Biology and ndhF sequence data. Evolution 56 (2), 233–252. Evolution 24, 1596–1599. Peng, Z., Ho, S.Y.W., Zhang, Y., He, S., 2006. Uplift of the Tibetan plateau: evidence Tao, J.R., Zhou, Z.K., Liu, Y.S., 2000. The Evolution of the Late Cretaceous–Cenozoic from divergence times of glyptosternoid catfishes. Molecular Phylogenetics and floras in China. Science Press, Beijing. Evolution 39, 568–572. Tapponnier, P., Zhiqin, X., Roger, F., Meyer, B., Arnaud, N., Wittlinger, G., Jingsui, Y., Peterson, A., Harpke, D., Peruzzi, L., Levichev, I., Tison, J.M., Peterson, J., 2009. 2001. Oblique stepwise rise and growth of the Tibet Plateau. Science 294, 1671– Hybridization drives speciation in Gagea (Liliaceae). Plant Systematics and 1677. Evolution 278, 133–148. Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmougin, F., Higgins, D.G., 1997. The Picard, D., Sempere, T., Plantard, O., 2008. Direction and timing of uplift propagation CLUSTAL_X windows interface: flexible strategies for multiple sequence in the Peruvian Andes deduced from molecular phylogenetics of highland alignment aided by quality analysis tools. Nucleic Acids Research 25, 4876– biotaxa. Earth and Planetary Science Letters 271, 326–336. 4882. Pinilla, J.C., Renner, S.S., Molecular Phylogenetics and Biogeography of Tiffney, B.H., 1985. Perspectives on the origin of the floristic similarity between Alstroemeriaceae, an Important clade of the Austral Floristic Realm. VW eastern Asia and eastern North America. Journal of the Arnold Arboretum 66, Status Symposium in Evolutionary Biology, 2010 Frauenchiemsee, Germany. 73–94. Qian, H., Ricklefs, R.E., 2000. Large-scale processes and the Asian bias in species U.S. Geological Survey, 2011. GEOLEX. . diversity of temperate plants. Nature 407, 180–182. Van Tuyl, J.M., Van Dijken, A., Chi, H.S., Lim, K.B., Villemoes, S., Van Kronenburg, Rabosky, D.L., 2006. LASER: a maximum likelihood toolkit for detecting temporal B.C.E., 2000. Breakthroughs in interspecific hybridization of lily. Acta shifts in diversifi cation rates from molecular phylogenies. Evolutionary Horticulture 508, 83–88. Bioinformatics Online 2, 247–250. Vinnersten, A., Bremer, K., 2001. Age and biogeography of major clades in Liliales. Rambaut, A., Drummond, A.J., 2007. Tracer v1.4. . American Journal of Botany 88, 1695–1703. Rešetnik, I., Liber, Z., Satovic, Z., Cigic´, P., Nikolic´, T., 2007. Molecular phylogeny and Wang, A., Yang, M., Liu, J., 2005. Molecular phylogeny, recent radiation and systematics of the Lilium carniolicum group (Liliaceae) based on nuclear ITS evolution of gross morphology of the rhubarb genus Rheum (Polygonaceae) sequences. Plant Systematics and Evolution 265, 45–58. inferred from chloroplast DNA trnL-F sequences. Annals of Botany 96, 489–498. Richardson, J.E., Pennington, R.T., Pennington, T.D., Hollingsworth, P.M., 2001a. Wang, H.W., Chen, J.M., Xu, C., Liu, X., Wang, Q.F., Motley, T.J., 2010. Population Rapid diversification of a species-rich genus of neotropical rain forest trees. genetic structure of an aquatic herb Batrachium bungei (Ranuculaceae) in the H- Science 293, 2242–2245. D Mountains of China. Aquatic Botany 92 (3), 221–225. Richardson, J.E., Weitz, F.M., Fay, M.F., Cronk, Q.C.B., Linder, H.P., Reeves, G., Chase, Wen, J., 1999. Evolution of eastern Asian and eastern North American disjunct M.W., 2001b. Rapid and recent origin of species richness in the Cape Flora of distributions in flowering plants. Annual Review of Ecology and Systematics 30, South Africa. Nature 412, 181–183. 421–455. Rieseberg, L.H., Soltis, D.E., 1991. Phylogenetic consequences of cytoplasmic gene Wendel, J.F. and J.J. Doyle. 1998. Phylogenetic incongruence: Window into genome flow in plants. Evolutionary Trends in Plants 5, 65–83. history and molecular evolution, In: Soltis D.E., Soltis P.S., Doyle J.J. (eds.), Rieseberg, L.H., Whitton, J., Linder, C.R., 1996. Molecular marker incongruence in Molecular Systematics of Plants II: DNA sequencing. Kluwer, Boston, plant hybrid zones and phylogenetic trees. Acta Botanica Neerlandica 45, 243– Massachusetts, pp. 265-296. 262. Wharton, P., Hine, B., Justice, D., 2005. The Jade Garden: New & Notable Plants from Ronquist, F., Huelsenbeck, J.P., 2003. MrBayes 3: Bayesian phylogenetic inference Asia. Timber Press, Portland. under mixed models. Bioinformatics 19 (12), 1572–1574. White, T.J., Bruns, T., Lee, S., Taylor, J., 1990. Amplification and direct sequencing of Rønsted, N., Law, S., Thornton, H., Fay, M.F., Chase, M.W., 2005. Molecular fungal ribosomal RNA genes for phylogenetics. In: Innis, M.A., Gelfand, D.H., phylogenetic evidence for the monophyly of Fritillaria and Lilium (Liliaceae; Shinsky, J.J., White, T.J. (Eds.), PCR Protocols: A Guide to Methods and Liliales) and the infrageneric classification of Fritillaria. Molecular Phylogenetics Applications. Academic Press, San Diego, pp. 315–322. and Evolution 35, 509–527. Willis, K.J., Whittaker, R.J., 2002. Species diversity-scale matters. Science 295, 1245– Royden, L.H., Burchfiel, B.C., Van der Hilst, R.D., 2008. The geological evolution of the 1248. Tibetan Plateau. Science 321 (5892), 1054–1058. Wilson, E.H., 1925. The Lilies of Eastern Asia: A Monograph. Dulau, London. Rundle, H.D., Nosil, P., 2005. Ecological speciation. Ecology Letters 8, 336–352. Wilson, E.O., 1992. The Diversity of Life. Harvard University Press, Cambridge, MA. Sang, T., Zhong, Y., 2000. Testing hybridization hypotheses based on incongruent Yang, Z.H., Rannala, B., 2006. Bayesian estimation of species divergencetimes under gene trees. Systematic Biology 49 (3), 422–434. a molecular clock using multiple fossil calibrations with soft bounds. Molecular Sang, T., Crawford, D.J., Stuessy, T.F., 1997. Chloroplast DNA phylogeny, reticulate Biology and Evolution 23, 212–226. evolution, and biogeography of Paeonia (Paeoniaceae). American Journal of Yu Y., Harris A., He X.J., in preparation. RASP (Reconstruct Ancestral State in Botany 84, 1120–1136. Phylogenies) 2.0 beta. Available from: . Bromham, L., Brown, G.K., Carpenter, R.J., Lee, D.M., Murphy, D.J., Sniderman, Zarrei, M., Wilkin, P., Fay, M.F., Ingrouille, M.J., Zarre, S., Chase, M.W., 2009. J.M.K., Udovicic, F., 2012. Testing the impact of calibration on molecular Molecular systematics of Gagea and Lloydia (Liliaceae; Liliales): implications of divergence times using a fossil-rich group: the case of Nothofagus (Fagales). analyses of nuclear ribosomal and plastid DNA sequences for infrageneric Systematic Biology 61, 289–313. classification. Annals of Botany 104, 125–142. Sealy, J.R., 1950. Nomocharis and Lilium. Kew Bulletin 5 (2), 273–297. Zhang, M.-L., Fritsch, P.W., 2010. Evolutionary response of Caragana (Fabaceae) to Sealy, J.R., 1983. A revision of the genus Nornocharis Franchet. Botanical Journal of Qinghai–Tibetan Plateau uplift and Asian interior aridification. Plant the Linnean Society 87, 285–323. Systematics and Evolution 288, 191–199. 460 Y.-D. Gao et al. / Molecular Phylogenetics and Evolution 68 (2013) 443–460

Zhang, D.F., Li, F.Q., Bian, J.M., 2000. Eco-environmental effects of the Qinghai–Tibet Zhang, Y.H., Volis, S., Sun, H., 2010. Chloroplast phylogeny and phylogeography of Plateau uplift during the Quaternary in China. Environmental Geology 39, Stellera chamaejasme on the Qinghai–Tibet Plateau and in adjacent regions. 1352–1358. Molecular Phylogenetics and Evolution 57, 1162–1172. Zhang, X.-L., Wang, Y.-J., Ge, X.-J., Yuan, Y.-M., Yang, H.-L., Liu, J.-Q., 2009. Molecular phylogeny and biogeography of Gentiana sect. Cruciata (Gentianaceae) based on four chloroplast DNA datasets. Taxon 58, 862–870.