Plant Syst. Evol. 246: 19–34 (2004) DOI 10.1007/s00606-003-0113-z

Phylogenetic relationships within the ‘core’ Laureae (Litsea complex, ) inferred from sequences of the chloroplast gene matK and nuclear ribosomal DNA ITS regions

Jie Li1, D. C. Christophel2, J. G. Conran3, and Hsi-Wen Li1

1Laboratory of Phylogenetics and Conservation Biology, Xishuangbanna Tropical Botanical Garden, The Chinese Academy of Sciences, Kunming, Yunnan, P.R. China 2Department of Biological Science, University of Denver, Iliff, Denver, CO, USA 3Centre for Evolutionary Biology and Biodiversity, Discipline of Environmental Biology, School of Earth and Environmental Sciences, Darling Building DP418, The University of Adelaide, Australia

Received April 17, 2003; accepted November 24, 2003 Published online: February 26, 2004 Springer-Verlag 2004

Abstract. A phylogenetic analysis of the ‘core’ Key words: Lauraceae, Laureae, matK, ITS, Laureae (Litsea complex) was conducted using the molecular phylogeny, Litsea, Lindera, classification. chloroplast gene matK and nuclear ribosomal DNA ITS sequences to investigate generic relationships The ‘core Laureae’ or Litsea complex consists and boundaries within the complex. Despite low genetic divergence for matK, rooting of the with of ten genera with 500–700 species concen- Sassafras resulted in Iteadaphne as the basal member trated in tropical-subtropical Asia, but with of the complex and five resolved clades: a Neolitsea representatives in Australasia, the Mediterra- clade and then Laurus, Parasassafras, Litsea and nean, North and Central America. The Lindera clades in a large polytomy with unresolved monophyly of the complex is well supported Lindera sections plus Umbellularia. A combined (Chanderbali et al. 2001; Rohwer et al. 1991; analysis of the data (identical to the ITS results) Rohwer 2000; Li 1985, 1995; van der Werff provided a more resolved phylogeny of the Laureae, and Richter 1996) and it is distinguished from with four major lineages: the Laurus, Litsea, Lindera other Lauraceae by possessing basically thyr- and Actinodaphne II clades. These clades also appear soid open, pseudo-racemose inflorescences to reflect the importance of inflorescence structure bearing involucral bracts and which appear and ontogeny within the Laureae, as well as data to be umbellate due to short axes, and mainly from cuticular micromorphology, but there was no support for traditional generic characters such as introrse anthers in the third whorl (Li and 2- versus 4- celled anthers. As a result, genera such as Christophel 2000, Li 2001). Actinodaphne, Litsea, Neolitsea and Lindera were Generic delimitation within the complex polyphyletic in all analyses. Parasassafras was has traditionally been problematic (Koster- related to Sinosassafras by the matK data, but mans 1957, Hutchinson 1964, Richter 1981), distant from it in the ITS and combined analyses. the major point being the significance of 20 J. Li et al.: Molecular phylogeny of the Litsea complex

2-locular versus 4-locular anthers as generic shaped fruiting cupules; sect. Daphinidium descriptors, although modern Lauraceae clas- (Nees) Benth. has evergreen, trinerved leaves sifications, tend to downplay the significance and un-developed terminal buds on shortened of anther cell number in generic delimitation brachyblasts; sect. Polyadenia Nees is ever- (e.g. Rohwer et al. 1991, Rohwer 1993, van der green with penninerved leaves and well-devel- Werff and Richter 1996). Li and Christophel oped terminal buds on shortened (2000) incorporated traditional morphological brachyblasts and sect. Uniumbellatae Tsui characters with leaf cuticles in a phylogenetic has evergreen, trinerved leaves, and the analysis of the complex, but although they also long-pedunculate pseudo- is solitary in concluded that anther cell number did not the axil of a normally developed leaf. have strong phylogenetic value at the generic In previous Lauraceae classifications, the level, they also demonstrated that morpholog- number of anther cells was a critical character ical data were insufficient to resolve relation- in distinguishing the different genera and ships within the Laureae. Kostermans (1957) considered it as the third The Litsea was divided into four most important character after inflorescence sections under the classification of Li et al. type and fruiting cupule shape. This view was (1984), with sect. Litsea defined as evergreen followed by Li (1985, 1995), Tsui (1987) and with alternate, penninerved leaves, a racemi- Hyland (1989), although Li (1985, 1995) con- form inflorescence and non-enlarged perianth sidered that the relationship may not be as tubes with absent or reduced perianth lobes. close as originally thought, and that the Sect. Conodaphne (Bl.) Benth. et Hook. f. is similarities between Litsea and Lindera may evergreen with alternate or opposite penni- be the result of parallel evolution. Li (1985) nerved leaves and non- to slightly enlarged and Tsui (1987) also suggested that there is perianth tubes. Sect. Cylicodaphne (Nees) parallel evolution between and within Litsea Hook. f. was considered to be evergreen with and Lindera at the generic and sectional levels, alternate, penninerved leaves, an enlarged making generic delimitation difficult using perianth tube and cup-shaped fruiting cupule. traditional morphological characters. Sect. Tomingodaphne (Bl.) Hook. f. has decid- Because morphological data did not pro- uous, alternate, penninerved leaves and non- vide strong phylogenetic signal, our study enlarged, 6-lobed perianth tubes. involving molecular sequences was undertaken Lindera was similarly divided into eight to find a more robust phylogeny for the sections by Tsui (1987) as follows: sect. Laureae. The matK gene is one of the most Lindera with deciduous, penninerved leaves rapidly evolving coding regions found so far in and well-developed terminal buds on short- the chloroplast genome (Olmstead and Palmer ened brachyblasts; sect. Sphaerocarpae Tsui 1994), and its high rates of synonymous and with deciduous; triplinerved or trinerved especially non-synonymous substitutions mean leaves and well-developed terminal buds on that it is considered appropriate for construct- shortened brachyblasts and sect. Palminervia ing infrafamilial phylogenies (e.g. Johnson and Meissn. with deciduous, lobed, trinerved Soltis 1994, 1995; Steele and Vilgalys 1994; leaves and well-developed terminal buds on Plunkett et al. 1996, 1997; Liang and Hilu shortened brachyblasts. Sect. Aperula (Bl.) 1996; Kron 1997; Gadek et al. 1996; Manos Benth. has evergreen, penninerved leaves, and Steele 1997; Xiang et al. 1998; Denda well-developed terminal buds on shortened et al. 1999), Nevertheless, Rohwer (2000) brachyblasts and long-pedunculate, racemi- found that matK sequence divergence in form inflorescences; Sect. Cupuliformis Tsui Lauraceae was surprisingly low (only 9.7% possesses evergreen, penninerved leaves, fun- informative characters), so a second marker nel-shaped glands on the third anther whorl was incorporated to improve recovery of and enlarged perianth tubes forming cup- phylogenetic signal (Chase et al. 1997). J. Li et al.: Molecular phylogeny of the Litsea complex 21

The ITS spacer region, has been utilised as in Actinodaphne in lateral sub-sessile fascicles’’. widely in phylogenetic studies of many taxa of van der Werff and Richter (1996) also considered flowering (e.g. Baldwin et al. 1995) and Sassafras to be close to the Laureae, noting that recent work on Lauraceae by Chanderbali both lack marginal parenchyma and possess similar et al. (2001) found that it provided good signal phloem fibres in the wood, nevertheless, it is resolution and therefore should be suitable for separated from the Laureae by its unique accentu- ated growth ring structure in both secondary xylem the Laureae. However, because ITS is a and phloem. nuclear gene spacer, phylogenetic conclusions Similarly, although Umbellularia is usually using ITS sequences are advised to be corrob- placed within the Laureae (Kostermans 1957, orated with data from other sources, including Rohwer 1993, van der Werff and Richter 1996), plastid genes such as matK (Donoghue and Chanderbali et al. (2001) and Rohwer (2000) Sanderson 1992). suggested that it should be placed in the Cinnamo- The goals of our study using maximum meae variously on the basis of ITS, trnL-F, parsimony analysis of sequence data from psbA-trnH and matK sequences and morphology nuclear ITS (internal transcribed spacer of (bisexual flowers and extrorse anthers in the ribosomal genes) and the chloroplast matK innermost staminal whorl). As a result, we have gene are to: (1) examine the circumscription of included it in our study, but treated it in the the Laureae; (2) test the hypothesis that Litsea, analyses as an outgroup. DNA extraction. Total DNA was extracted Lindera, Actinodaphne and Neolitsea are each using the CTAB method of Doyle and Doyle monophyletic; and (3) investigate major lin- (1987). Approximately 1 cm2 leaf tissue was ground eages in order to try to explain evolution with a minute amount of sterile sand and 1 ml of within the complex. warmed (60 C) CTAB + BME in a warmed (60 C) mortar and pestle; an additional 1 ml CTAB was added and mixed, then the slurry was Materials and methods poured into two 1.5 ml tubes and incubated at Ingroup sampling. Taxon sampling included repre- 60 C in a water bath for 45 minutes. Two volumes sentatives of all previously recognised generic of chloroform: isopropanol (24:1) were added, the segregates in the complex, as well as sections of tubes rocked for at least 10 minutes and then larger genera such as Litsea and Lindera as defined microfuged for 10 minutes at 14,000 rpm. The by Nees von Esenbeck (1836), Meissner (1864), aqueous phase was transferred to a fresh tube to Bentham (1880), Pax (1889), Hooker (1890), Ko- which 1 ml cold 100% EtOH was added and stermans (1957), Hutchinson (1964), Li (1984, incubated at )20 C for between 30 minutes and 1985), Long (1984) and Tsui (1987). A complete overnight to precipitate DNA. The tube was then list of the species sampled, along with collection centrifuged for 10 minutes at 14,000 rpm, the and voucher information is provided in Table 1. supernatant decanted, and the pellet washed for 5 Outgroup selection. Sassafras is generally con- minutes using 500 ll wash buffer (70% EtOH, sidered to be closely related to the Laureae due to l0 mM NH4OAc). After centrifugation for another its dioecious breeding system, inflorescences with 10 minutes at 14,000 rpm, the supernatant was involucral bracts and introrsely positioned locelli in again decanted, and the pellet dried and then all staminal whorls, (Kostermans 1957, Rohwer resuspended in 100 ll ddH2O (50 ll for herbarium 1993), as well as being sister to the clade in the ITS specimens). study of Chanderbali et al. (2001). However, as Where DNA was difficult to extract by this Rehder (1920) noted: ‘‘From all the genera of the method, the QIAquick spin column was used to tribe Litseae the genus Sassafras is easily separated extract DNA (M. Chase, pers. comm.), 150 llof by its racemes of slender pedicellate flowers in the the aqueous phase mentioned above was added to axils of the basal scales of the terminal branch-bud, the column, after which the QIAquick protocol while in other genera the flowers are arranged in for cleaning PCR products was followed. This lateral or heads, sometimes reduced to 1 method avoided precipitating the DNA and gener- flower, subtended by an involucre of 4–6 bracts, or ally produced good results within a few minutes. 22 J. Li et al.: Molecular phylogeny of the Litsea complex

Table 1. Species of the Laureae and outgroup taxa included in the analysis Taxon Voucher Source GB number GIB number for matK for ITS Litsea section Tomingodaphne Li H.-W. 28 Yunnan, China AF 244398 AY265402 Litsea cubeba (Lour.) Pers. (HITBC) Section Litsea Li H.-W. 21 Yunnan, China AF 244396 AY265403 Litsea glutinosa (HITBC) (Lour.) C.B. Rob. Section Conodaphne Li H.-W. 24 Yunnan, China AF 244395 AY265404 Litsea umbellata (Lour.) Merr. (HITBC) Section Cylicodaphne Li H.-W. 19 Yunnan, China AF 244397 AY265405 Litsea dilleniifolia P.Y. Pai (HITBC) et P.H. Huang Lindera Section Cupuliformes Li H.-W. 7 Yunnan, China AF 244404 AY265406 Lindera megaphylla Hemsl. (HITBC) Section Lindera Nei M.X. & Jiangxi, China AF 244401 AY’265407 Lindera reflexa Hemsl. Lai S,.K. 3768 (KUN 0100201) Section Aperula Li H.-W. 8 (HITBC) Yunnan, China AF 244403 AY265408 Lindera metcalifiana Allen Li H.-W. 4 (HITBC) Section Polyadenia Li G.F. 4 (HITBC) Yunnan, China AF 244406 AY265409 Lindera communis Hemsl. Section Sphaerocarpae Li G.F. 63966 Sichuan, China AF 244405 AY265410 Lindera fruticosa Hemsl. (KUN 0104915) Section Palminerviae Sino-Amer. Hubei, China AF 244402 AY265411 Lindera obtusiloba Bl. Exped. 1308 (KUN 0151469) Section Uniumbellatae Sichuan Exped. 2258 Sichuan, China AF 244399 AY265412 Lindera tienchuanensis (KUN0101388) W.P. Fang et H.S. Kung Section Daphnidium Li H.-W. 9 Yunnan, China AF 244400 AY265413 Lindera thomsonii Allen (HITBC) Actinodaphne I: Actinodaphne obovata Li H.-W. 1 Yunnan, China AF 244410 AY265398 (Nees) Bl. (HITBC) II: Actinodaphne forrestii Li H.-W. 2 Yunnan, China AF 244411 AY265399 (Allen) Kosterm. (HITBC) Neolitsea I: Neolitsea confertifolia GaoX.P. 53971 Guangdong, AF 244394 AY265400 (Hemsl.) Merr. (KUN 0106507) China II: Neolitsea levinei Merr. Li H.-W. 29 Yunnan, China AF 244393 AY265401 (HITBC) Iteadaphne Iteadaphne caudata (Nees) Li H.-W. 27 Yunnan, China AF 244408 AY26.5396 H.-W. Li (HITBC) Dodecadenia Dodecadenia grandiflora Nees Wu C.Y. et al Tibet, China AF 244409 AY265397 75-1048 (KUN 0049206) J. Li et al.: Molecular phylogeny of the Litsea complex 23

Table 1 (continued) Taxon Voucher Source GB number GIB number for matK for ITS Parasassafras Parasassqfras confertiflora Qian Y.Y. 682 Yunnan, China AF 244392 AY265395 (Meissn.) Long (KUN 0104558) Sinosassafras Sinosassafras flavinervia Yang Z.H. 101437 Yunnan, China AF 244390 AY265394 (Allen) H.-W. Li (KUN 0150376) Laurus Laurus nobilis L. Li H.-W. 16 (HITBC) Yunnan, China AF 244407 AY265392 Outgroup taxa: Sassafras tzumu Li H.-W. 15 (HITBC) Yunnan, China AF 244391 AY265391 (Hemsl.) Hemsl. Umbellularia californica van der Werff s.n. North America AF 244389 AY265393 (Hooker et Arnott) Nuttall (MO)

PCR and sequencing. The main primers used 5300, 5200, 5200F, 1245, 4100, 5800 and 2288 to amplify the whole matK gene were 909 and 2288 (Table 2). The entire ITS region was amplified (Johnson and Soltis 1995). However, if these were successfully in most cases using LAUR 1 and ITSB unsuccessful, various internal primer combinations (Chanderbali et al. 2001), but if these failed, primer were utilised using the primers 909, 5500, 5400, combinations using the universal primers of White

Table 2. Primers used for amplification and sequencing matK and ITS in the Laureae matK Primer sequence 5¢-3¢ Source Forward 909(trnK3914F) GGGGTTGCTAACTCAACGG Johnson et al. (1996) 1244(matK7) GTATTAGGGCATCCCATT Steele and Vilgalys (1994) 1245(KPS5) GGATCCTTTCATGCATTATG Steele and Vilgalys (1994) 5400 CTCAAATGATATCGAAGGG L. Jones unpubl. Primer 5500 GATGGATTTCGGCAACAATA L. Jones unpubl. primer 5600 GTGTACGACTAAACTCTTCG L, Jones unpubl. primer 5200F CCTTTCTTGAGCGAACAC L. Jones unpubl. primer Reverse 2288 (trnK-2R) AACTAGTCGGATGGAGTAG Johnson et al. (1996) 4100 TAGAACTAGATAGATCTCAGC K. Edwards unpubl. primer 2520 GATCCTTCCTGGTTGAAACCAC L. Jones unpubl. primer 5200 GTGTTCGCTCAAGAAAGG L. Jones unpubl. primer 5300 GCATCTTGTATCCAAGAGTG L. Jones unpubl. primer 5800 GGTTCTCTATGTGACCTATG L. Jones unpubl. primer ITS Primer sequence 5¢-3¢ Source Forward LAUR 1 ACCACCACCGGCGAACCA Chanderbali et al. (2001) ITS3 GCATCGATGAAGAACGTAGC White et al. (1990) USA GGAAGGAGAAGTCGTAACAAGG Blattner (1999) Reverse 1TS2 GCTACGTTCTTCATCGATGC White et al. (1990) ITSB CTTTTCCTCCGCTTATTGATATG Blattner (1999) ITS4 TCCTCCGCTTATTGATATGC White et al. (1990) 24 J. Li et al.: Molecular phylogeny of the Litsea complex et al. (1990) with LAUR 1 and ITS B were also branch-swapping, and characters re-weighted using used to amplify regions of poor-quality template. their rescaled consistency indices and the maximum The primer sequences used are shown in Table 2. value (best fit) criterion. All from these 10 To prevent PCR contamination, a negative replicates were then swapped to completion, after control was used for every primer combination in which re-weighting was repeated until tree length every reaction. Once the required fragment was stabilised. Homoplasy was estimated by consistency successfully amplified, the DNA was purified using (excluding uninformative sites) and retention indi- the Qiagen QIAquick PCR Purification Kit, ces. Internal support was evaluated using Bootstrap following protocols provided by the manufacturers. analysis (Felsenstein 1985), with 1000 replicates To sequence the DNA fragment, a series of performed on the weighted data matrix, using the reactions were set up, each with one different following PAUP* settings: retain groups compatible internal primer (Table 2), thus producing single with 50% majority-rule consensus, sample charac- stranded DNA in both directions to produce ters with equal probability but apply weights, 10 overlapping sequences. Sequencing reactions were replicate random sequence addition, TBR branch- performed in a 0.2 ml PCR tube, using 2 llof swapping and MULPARS off. primer, 90 ng of DNA template, 4 ll of Big Dye and 4 ll of Big Dye Buffer, using ddH2O to make up a final reaction volume of 20 ll. This mixture Results was then temperature cycled at 1) 96 C for 30 Sequence characteristics. DNA was extracted seconds, 2) 50 C for 15 seconds, 3) 60 C for 4 minutes with 25 cycles. Cleaned products were then successfully from all 23 samples, including eight directly sequenced in an Applied Biosystems 3100 from herbarium samples more than 20 years DNA automated sequencer. old. The amplified matK regions were 1406 base Sequence alignment. Sequences of the 23 taxa pairs (bp) long, except for those of Lindera were aligned using DAPSA ver. 3.8 (Harley 1995), tienchuanenis W. P. Fang et H. P. Kung from allowing uncertainties either to be resolved or Lindera sect. Uniumbellatae and Lindera recorded as ambiguities. When all the overlapping thomsonii Allen from Lindera sect. Daphnidium sequences had been checked, a consensus sequence which were missing 108 bp and 31 bp, respec- for each species was generated and aligned to the tively (possibly due to primer mismatch or poor- database sequences and any differences reverified. quality templates). With no insertions or Consensus sequences for all taxa were then deletions alignment posed no problem, however, realigned and a final data matrix produced. as matK is highly conservative in Lauraceae Phylogenetic analysis. The data matrix for the 23 taxa consisted of a matK submatrix comprising (Rohwer 2000) the number of nucleotide sub- 1406 base pairs (bp) without alignment gaps in the stitutions was limited. Informative sites com- coding region, and an ITS submatrix comprising prised only 1.14% of 1,406 sites (16 bp), and the 673 bp with 16 alignment gaps. Only phylogenet- Guanine and Cytosine (GC) content ranged ically informative characters were analysed, gaps from. 36.1–36.6%, with an average of 36.3%. were scored as missing data, and all characters were In contrast, the ITS region was quite unweighted initially and treated as unordered variable between taxa, and alignment required multistate. Resulting trees were rooted by outgroup 16 indels varying from. 1–44 bp. As a result, comparison with Sassafras and Umbellularia. only regions which could be aligned unequiv- Parsimony analysis was performed using * ocally were analysed, making our ITS phylo- PAUP version 4.0b10 (Swofford 1998), analysing genetic estimates conservative. These aligned the aligned submatrices alone and then combined. sequences included the 5.8S gene and resulted An initial heuristic search with 100 replicates with random addition sequence, tree-bisection-reconnec- in 299 constant, 426 uninformative, and 247 tion (TBR) branch-swapping, MULPARS on, and parsimony informative characters. The total steepest descent off, was followed by successive aligned ITS region contained 673 bases with weighting (Farris 1989), using heuristic searches with 36.7% parsimony informative characters, 10 replicates and random sequence addition, TBR although due to the poor quality of DNA J. Li et al.: Molecular phylogeny of the Litsea complex 25 from some herbarium specimens, data for matK analysis. Analysis of the matK ITS1 were missing in Lindera tienchuanensis. submatrix resulted in 45 equally parsimonious The GC content of ITS was much higher than trees (length = 19; CI = 0.8947; RI = matK, ranging from 51.0% in Neolitsea con- 0.9048). The results from the successive weight fertifolia (Helms.) Merr. to 67.5% in Litsea analyses were identical to those of the pre- glutinosa (Lour.) C.B. Rob. weighted analyses, and the topology of the The combined data matrix therefore con- 50% majority rule consensus tree is shown in tained 23 OTUs and 2,079 characters, of which Fig. 1. 263 were parsimony informative.

Fig. 1. The majority consensus tree of 45 equally parsimonious trees obtained from the successive weighting analysis of the matK submatrix. Majority rule percentages are indicated above branches and bootstrap support below branches 26 J. Li et al.: Molecular phylogeny of the Litsea complex

The resulting cladogram was poorly re- forming a terminal pair clade with relatively solved with Actinodaphne, Lindera and Litsea low bootstrap support (64%), which was sister all polyphyletic, and only five small terminal to Lindera section Sphaerocarpae (51% sup- clades common to all 45 trees. The first of port). these (A) represented the two sampled Neolit- Actinodaphne I, Iteadaphne, Lindera section sea accessions (97% bootstrap support) weakly Aperula, Dodecadenia, Litsea sections Litsea supported as sister to Actinodaphne I, the and Cylicodaphne constituted Clade C, with second (B) a Laurus – Dodecania pair with Actinodaphne I sister to the remainder. Litsea 89% bootstrap as unsupported sisters to a sections Litsea and Cylicodaphne constituted a Litsea sects. Conodaphne and Cylicodaphne monophyletic pair with 73% bootstrap sup- pair. Clade C contained Sinoassafras and port and allied with Dodecadenia (78%). Lin- Parasassafras with 100% bootstrap support dera section Aperula was sister to that trio, but whereas the fourth clade (D) consisted of with only weak support (57%). Actinodaphne II and Lindera sect. Cupiliformes In Clade D, Sinosassafras, Neolitsea I and (67% support) along with Litsea sects. Litsea Lindera section Uniumbellatae formed an and Daphnidium. The final clade (E) repre- unresolved but well-supported polytomy sented a polytomy of three Lindera sections (100% bootstrap support), which formed a (Lindera, Sphaerocarpae and Tomingodaphne) strongly supported (100%) sister group to but only poor support (62%). Rooting the tree Lindera section Daphnidium. Neolitsea II was to Sassafras meant that Umbellularia weakly allied to them (61% support), and (although a nominated outgroup) fell within Laurus was basal to the clade, but without the Laureae as part of the large unresolved bootstrap support. polytomy, but without bootstrap support. Combined matK and ITS. Heuristic analy- ITS analysis. Analysis of the ITS se- sis of all phylogenetically informative nucleo- quences resulted in 15 most parsimonious trees tide characters from the combined chloroplast (length = 575; CI = 0.6330; RI = 0.6719). and nuclear DNA data produced 15 trees each Successive weighting produced 3 shortest trees 611 steps long (CI = 0.6236; RI = 0.6536). (length = 364; CI = 0.7482, RI = 0.7914), all Successive weighting resulted in three most of which were congruent with the trees from parsimonious trees, each 381 steps long (CI = the unweighted analysis, and the topology of 0.7373; RI = 0.7773) (Fig. 2). the weighted strict consensus tree is shown in The strict consensus tree from the com- Fig. 2. bined analysis is topologically identical to the Four major clades occur in the strict ITS submatrix analysis, the same clades reap- consensus tree, although none of these had pearing in the combined analysis, and with bootstrap support. The first and most basal of similar bootstrap support values. However, these (Clade A) contained Lindera section only one of the bootstrap-supported clades Palminerviae, Litsea section Tomingodaphne, from the matK analysis clades (Actinodaphne Lindera section Cupuliformes, and Actino- II + Lindera section Cupuliformes) occurred in daphne II. Within the clade, Actinodaphne II the combined matrix. and Lindera section Cupuliformes formed a terminal pair with 100% bootstrap support, Discussion and sister to Litsea section Tomingodaphne with 74% bootstrap support. Lindera section Data combinability. The combinability of dif- Palminerviae was basal within Clade A, but ferent data sets is a contentious issue and without bootstrap support. there are numerous suggested methods for Clade B comprised Parasassafras, Lindera approaching the problem (e.g. De Queiroz section Sphaerocarpae, Lindera section Lindera et al. 1995, Huelsenbeck et al. 1996, Nixon and Litsea section Conodaphne, the latter two and Carpenter 1996), although if incongruence J. Li et al.: Molecular phylogeny of the Litsea complex 27

Fig. 2. The strict consensus tree of the three equally most parsimonious trees resulting from the successive weighting analysis of a combined matK and ITS matrix, and identical to the topology of the ITS only analysis. Bootstrap support percentages for the combined analysis are indicated above the branches, those for ITS only, below the branches is low, or there is relatively low (<70%) As there was mostly lower bootstrap support bootstrap support for the incongruent nodes in at least one of the two trees for the then data can be deemed to be combinable incongruent clades, our Laureae data were (e.g. Mason-Gamer and Kellogg 1996, El- considered to be sufficiently compatible to be dena¨s and Linder 2000, Barker et al. 2003). combined. Despite the small number of 28 J. Li et al.: Molecular phylogeny of the Litsea complex parsimony informative sites found in matK cpDNA sequences also found that Umbellu- for the Laureae (1.14% of the total) com- laria was part of what they called the Ocotea pared to ITS (36.7%), limiting its usefulness complex, with similar relationships to Roh- for classification in our study, it showed low wer’s matK results, and that it was not part of levels of homoplasy. ITS had a greater the Laureae. It is possible that in the absence number of parsimony informative sites, but of its near relatives, Umbellularia fell inside the region was also more homoplasious, Laureae in our matK analysis due to the very resulting in lower indices of consistency and small number of informative sites and/or long- retention. branch attraction. When the matK and ITS sequences were These features, and the results of our combined, the results were identical to those combined study tend to support the exclusion obtained with ITS alone, probably because of both genera from the Laureae, rather than ITS had more than thirty times the number of the inclusion of at one of the two, depending informative sites, and combining the two did on which is chosen as the outgroup, as our not substantially alter the levels of bootstrap matK data might suggest. support for the nodes. Combined data sets Polyphyly in Litsea and Lindera. Hooker generally produce more robust phylogenies (1890), although following Bentham (1880), (Chase et al. 1997, Davis et al. 1998), espe- expressed similar doubts as to the utility of cially where nuclear and chloroplast genes may 2- and 4-celled anthers as generic characters, have different histories due to mutation, intro- and Rohwer (1993) considered that Litsea and gression, transfer into different lineages or loss Lindera were anatomically and morphologi- of polymorphisms in descendant species (Judd cally polyphyletic, and that they should be split et al. 1999) Nevertheless, caution is still rec- into smaller entities. The representatives of the ommended, and conflict between data sets sections of these genera sampled were poly- should at least be taken into consideration. phyletic in our study, agreeing with the Circumscription of the Laureae. The inclu- morphological and leaf cuticle analysis of Li sion of Sassafras and/or Umbellularia within and Christophel (2000), suggesting that these the Laureae had been advocated by several genera are not monophyletic. This further authors (Kostermans 1957, Rohwer 1993, van supports the assertions by Hyland (1989), der Werff and Richter 1996) based on intrors- Rohwer et al. (1991), van der Werff and ely positioned anther locelli and/or the um- Richter (1996) and van der Werff (2001) that bellulate, involucrate inflorescences. However, the use of 2-thecate and 4-thecate anthers in in the Laureae introrse locelli occur only in the generic delimitation was confusing and likely outer two staminal whorls (mainly in the to result in the recognition of poly- or third), not in all whorls as in Sassafras, and paraphyletic genera. the inflorescences in Sassafras are botryose not Nevertheless, until a much wider-sampled umbellulate. and robust phylogeny can be developed for the Similarly, despite possessing umbellulate, tribe, the poly- or paraphyly of both these involucrate inflorescences, Umbellularia has genera, and the relationships and possibly even bisexual flowers and extrorse locelli in the monophyly of the sections within them will innermost staminal whorl. Although Umbellu- remain unresolved. laria fell inside Laureae in our matK analysis Systematic position of Sinosassafras, Para- (if the tree is rooted to Sassafras) the more sassafras, Iteadaphne and Dodecadenia. As extensive family-level matK study by Rohwer with the Litsea/Lindera complex, two other (2000) found that the former was instead generic pairs traditionally distinguished by associated with Licaria and Endlicheria in a anther cell number (Sinosassafras vs. Parasas- terminal clade with members of Cinnamo- safras and Iteadaphne vs. Dodecadenia) were meae. Chanderbali et al. (2001) using ITS and also unrelated in our ITS and combined J. Li et al.: Molecular phylogeny of the Litsea complex 29 analyses. Nevertheless, Sinosassafras and The apparent polyphyly of Actinodaphne is not Parasassafras were monophyletic with 100% surprising, given the absence of any clear bootstrap support in our matK analysis, morphological synapomorphies for the genus, agreeing with Rohwer’s (1993) assertion that but wider sampling is needed to confirm this. anther locule number was an insufficient In Neolitsea, at least based on the very ground for generic separation. Although there limited sample used in our study, the genus were very few informative matK sites, the fact divided into pinnately-veined (Neolitsea I) and that this pair was strongly supported in our tri-nerved (Neolitsea II) entities, which is also data suggests that given the more conservative consistent with both gross morphology and nature of matK their pairing may well reflect leaf cuticular features (Li and Christophel phylogenetic signal. 2000). The traditional recognition of Neolitsea In contrast, although Iteadaphne and Do- based on its dimerous flowers was neither decadenia have umbels which are reduced to a supported by ITS nor the combined analysis. single flower, the different number and Nevertheless, the genus was a monophyletic arrangement of involucral bracts caused Roh- clade in the matK analysis with very high wer (1993) to treat them as separated genera, bootstrap support (97%), and as with the and this is supported by our study. Neverthe- Sinosassafras/Parasassafras matK pair, the less, they, along with the core Litsea group and more conservative nature of that gene may be part of Actinodaphne formed the unsupported indicating phylogenetic signal which requires Litsea clade, partly reflecting Kosterman’s further investigation with more robust sam- (1957) placement of Dodecadenia within Lit- pling. sea, but also Rohwer’s (1993) suggestion that Phylogenetic relationships within the Lau- Dodecadenia was possibly close to Actin- reae. The results of our study and the molec- odaphne. Our matK analysis, on the other ular phylogenies of Chanderbali et al. (2001) hand, placed Dodecadenia in a well supported and Rohwer (2000), show that traditional clade with Laurus again suggesting that much diagnostic generic and sectional characters wider sampling is needed in order to determine such as deciduousness, number of leaf nerves, fully the monophyly and relationships between the size and arrangement of involucral bracts, Laureae genera and sections. the numbers of floral parts and flowers per Actinodaphne and Neolitsea. Imbricate, pseudo-umbel, number of anther cells, shape deciduous involucral bracts at the base of the of the cupule etc… have arisen repeatedly inflorescence traditionally delineate Actin- within the family. Accordingly, the homology odaphne (Nees von Esenbeck 1836). However, of such characters needs to be reconsidered in the genus was polyphyletic in both the matK light of the molecular data, as many appear to or ITS analyses, agreeing with the morpholog- be homoplasious. The convergent nature of ical study of Li and Christophel (2000). these characters is evident not only from their Our study suggests that Actinodaphne distributions within polyphyletic genera, but might be separated into two groups, one also because they occur several times in characterised by thyrsoid inflorescences (Ac- unrelated genera. This means that although tinodaphne I); the other with clustered or some characters appear to support the molec- fasciculate pseudo-umbels (Actinodaphne II), ular clades, the higher-level taxonomic utility and this separation is also supported by leaf of morphological data in Laureae must be micromorphology, where Actinodaphne I is treated with caution, until it can be determined mainly characterised by granulate periclinal through more extensive sampling and delinea- walls on the abaxial epidermis, but Actino- tion of robust clades, which, if any, morpho- daphne II has finely papillate periclinal walls logical characters provide enough resolution to with narrow or linear stomatal scales on the reflect relationships at the level of genus and/or abaxial epidermis (Li et al. unpublished data). section. 30 J. Li et al.: Molecular phylogeny of the Litsea complex

Based on the molecular results, it was also studies of Litsea and Lindera, which we here clear that generic boundaries within the com- term the Brachyblast Type, and which is plex seem to require extensive redefinition, but exemplified by the Laurus Clade (D). The using traditionally morphology it is difficult to ancestral thrysoid inflorescence is reduced to find synapomorphies which define these molec- a pseudo-umbellate one, which then con- ular clades. Nevertheless, mapping of some of denses to multiple shortened brachyblasts the more frequently used generic and supra- and, by reduction, to solitary apparently generic characters in Lauraceae by van der racemose pseudo-umbels. Elongation of the Werff and Richter (1996) suggested that inflo- axis could then lead to the condition of a rescence features were among the most reli- solitary, axillary, long-pedunculate pseudo- able. umbel, a condition which may have arisen According to Li (1985, 1995) and Tsui several times. (1987), the evolutionary trends in Laureae, A second inflorescence development series and Litsea and Lindera in particular, are as noted by Rohwer (1993) can be seen in the follows: evergreen to deciduous, pinnate to Litsea Clade (C) where a thyrsoid cymose triple-nerved or trinerved leaf venation, 3- inflorescence is reduced by shortening of merous to 2-merous flowers, and the evolu- brachyblast internodes to the pseudo-racemose tionary trend of progressive reduction in the arrangement seen in Lindera section Aperula, inflorescence, each section representing a Litsea sections Litsea and Cylicodaphne. In the different stage of inflorescence development, final stage the number of flowers per involucre and this is largely reflected in the ITS is reduced to one; the apparently convergent sequence-derived topology (Chanderbali et al. condition seen in both Iteadaphne and Do- 2001). In our study, each of the clades in the decadenia. combined analysis is homogenous, but each The third type of inflorescence develop- clade represents different characters: some ment seen in the Lindera Clade (B), is where a have 2-merous flowers, some have trinerved pair of pseudo-umbels occurs on each side of a leaves, and some others show different stages terminal bud which then develops into a leafy of inflorescence evolution. shoot. In contrast, the alternative clustered Kostermans’ (1957) regarded Sassafras of inflorescence form seen in the Actinodaphm II tribe Cinnomomeae to be part of the ancestral Clade (A) consists of several pseudo-umbels lineage leading to the Laureae, but recent clustered around a terminal bud from which molecular data (Rohwer 2000, Chanderbali either a new leafy shoot or several pseudo- et al. 2001) showed that Cinnamomeae umbels and a terminal bud develops, resulting (including both Sassafras and Umbellularia) in a mixed bud enclosed by imbricate scales. are sister to the Laureae, hence our use of them Tsui (1987) considered that these clustered as outgroup taxa. inflorescences also resulted from shortening of The Laureae are characterised by basi- the brachyblast, resulting in the two or more cally thyrsoid inflorescences that are often pseudo-umbels making the terminal bud ap- protected by decussate or alternate bracts. pear to be axillary. Although there is general agreement that However, the lack of bootstrap support pseudo-umbels result from shortening of the for the clades in our combined analyses as inflorescence axis (Li 1985, Tsui 1987, Roh- well as the small sample size relative to the wer 1993, van der Werff and Richter 1996), numbers of species in the tribe as a whole there are relatively few studies of inflores- make statements about character evolution cence evolution within the Lauraceae and highly speculative. Nevertheless, the fact that Laureae in particular (Weberling 1985). Li there are apparent developmental patterns (1985) and Tsui (1987) suggested an evolu- suggests that inflorescence development may tionary series for the inflorescence based on well be one of the more important sources of J. Li et al.: Molecular phylogeny of the Litsea complex 31 morphology-based phylogenetic signal in Conclusions Laureae. Certainly inflorescence features were Phylogenetic analysis of the ‘core’ Laureae regarded by Rohwer (1993) as being impor- (Litsea complex) using matK and ITS se- tant in discussing generic affinities, and they quences provided a relatively resolved but very were also seen in broad terms (paniculate poorly bootstrap-supported phylogeny of the versus thyrsoid versus involucrate) to map Laureae, with genera such as Actinodaphne, well onto the major lineages within the family Litsea, Neolitsea and Lindera polyphyletic in (Rohwer 2000). Whether the finer scale devel- the analyses. Four major clades resulted which opmental series within the Laureae hold up are referred as the Laurus, Litsea, Lindera and under broader, more robust sampling remains Actinodaphne II clades, based on the placement to be seen, but these patterns merit further of the various type sections of the genera investigation. within each group. These clades appear to Biogeography. Li (1995) inferred on the reflect the importance of inflorescence struc- basis of the large number of endemic genera ture and ontogeny within the Laureae, as well and species found in tropical and subtropical as data from cuticular micromorphology, but Asia that the Laureae possibly originated in there was no support for traditional generic the southern part of Laurasia or northern characters such as 2- versus 4-celled anthers. part of Gondwana along the tropical coast However, because of the apparent fragmen- area of the Tethys Sea not earlier than the tation of these large genera based on relatively mid-Cretaceous. Particularly for Litsea and small taxon samples, as well as conflict between Lindera, which are the core genera in the some well-supported clades in the matK tree complex, the area around South China to versus the combined analysis, much more Indo-Malaysia was proposed as the centre of detailed studies are required to clarify the origin and speciation, with later migration relationships which emerged in our study and into tropical America and Australasia. The to allow for more precise generic boundary fossil record for Australasia shows that unlike definitions and hypotheses about the historical earlier theories of recent invasion (Barlow biogeography of the tribe. Similarly, the phy- 1981), the families predate the breakup of logenetic significance of inflorescence and other Gondwana and are part of the autochthonous morphological characters needs to be deter- Australian rainforest flora (Christophel 1994), mined in the light of these new alignments. supporting an early widespread Gondwanan Our study also highlighted several areas in distribution. Whereas Rohwer (2000) found need of further study. More data, including that the basal Lauraceae (Cryptocaryeae) both more characters and more taxa, are were Gondwanan, the Laureae were part of required before a well-resolved robust phylog- a terminal Laurasian/South American group eny can be produced. In particular, extensive (although including taxa from Australia and sampling is needed within the large polyphy- New Zealand). Similarly, Chanderbali et al. letic such as Litsea, Lindera and Neolitsea. (2001) placed the Laureae within a boreo- tropical group, and in both studies the Laurasian Perseae were basal to the Lau- Jens Rohwer, Henk van der Werff and Paul reae/Cinamomeae clades. This, combined Gadek are thanked for valuable suggestions and generously providing leaf material. Thanks are also with the absence of any clear Gondwanan due to Lynne Jones, Karen Edwards, Maggie versus Laurasian clades emerging in our study Serewko, Qi-sen Zhang and Matthew Donnon for tends to support the idea of a Laurasian assistance with sequencing. The Directors of HIT- origin but with early and perhaps repeated BC, KUN, QRS, PE, SING, TAIF, BO and AD migration and diversification into Gondwana, are thanked for the loan and storage of specimens. although more extensive sampling may help This study had its origin partly in Jie Li’s to improve the picture. doctoral dissertation at the then Department of 32 J. Li et al.: Molecular phylogeny of the Litsea complex

Environmental Biology, The University of Adela- Davis J. I., Simmons M. P., Stevenson D. W., ide, South Australia, which is thanked for the Wendel J. F. (1998) Data decisiveness, data provision of resources. The work was also sup- quality, and incongruence in phylogenetic anal- ported by the grants from NSFC (30200017), ysis – an example from the monocotyledons YNSF (2001C0008R), CASTALENTS using mitochondrial atpA sequences. Syst. Biol. (20010713093959), CEF (498) and OPRS to J. Li. 47: 282–310. De Queiroz A., Donoghue M. J., Kim J. (1995) Separate versus combined analysis of phyloge- References netic evidence. Ann. Rev. Ecol. Syst. 26: 657–681. Denda T., Watanabe K., Koosuge K., YaharaT., Ito Baldwin B. G., Sanderson M. J., Porter J. M., M. (1999) Molecular phylogeny of Brachycome Wojciechowski M. F., Campbell C. S., Don- (Asteraceae). Plant Syst. Evol. 217: 299–311. oghue M. J., Soltis P. S., Kuzoff R. K., Ko S. C., Donoghue M. J., Sanderson M. J. (1992) The O’Kane S. L., Jr., Schaal B. A., Liu Z. L., Sinclair suitability of molecular and morphological evi- J. B., Fangan B. M., Stedje B., Stabbetorp O. E., dence in reconstructing plant phylogeny. In: Jensen E. S., Jakobsen K. S., Mes T. H. M., t’Hart Soltis P. S., Soltis D. E., Doyle J. J. (eds.) H. T. (1995) The ITS region of nuclear ribosomal Molecular systematics of plants. Chapman and DNA: A valuable source of evidence on angio- Hall, New York, pp. 340–368. sperm phylogeny. Ann. Missouri Bot. Gard. 82: Doyle J. J., Doyle J. S. (1987) A rapid DNA 247–277. isolation procedure for small qualities of fresh Barker N. P., Linder H. P., Morton C. M., Lyle M. leaf tissue. Phytochemical Bull. 19: 11–15. (2003) The paraphyly of Cortaderia (Dantho- Eldena¨s P., Linder H. P. (2000) Congruence and nioideae; Poaceae): evidence from morphology complimentarity of morphological and trnL-F and chloroplast and nuclear DNA sequence sequence, and the phylogeny of the African data. Ann. Missouri Bot. Gard. 90: 1–24. Restionaceae. Syst. Bot. 25: 692–707. Barlow B. A. (1981) The Australian Flora: Its Farris J. S. (1989) A successive approximations origin and evolution. In: George A. S. (ed.) approach to character weighting. Syst. Zool. 18: Flora of Australia Vol. 1. Australian Govern- 374–385. ment Publishing Service, Canberra, pp. 25–73. Felsenstein J. (1985) Confidence limits on phylog- Bentham G. (1880) Laurineae. In: Bentham G., enies: an approach using the bootstrap. Evolu- Hooker J. D. (eds.) Genera Plantarum, Vol. 3. tion 39: 783–791. London, L. Reeve, pp. 146–168. Gadek P. A., Wilson P. G., Quinn C. J. (1996) Blattner F. R. (1999) Direct amplification of the Phylogenetic reconstruction in Myrtaceae using entire ITS region from poorly preserved plant matK, with particular reference to the position material using recombinant PCR. Biotechniques of Psiloxylon and Heteropyxis. Austral. Syst. 27: 1180–1186. Bot. 9: 283–290. Chanderbali A. S., van der Werff H., Renner S. S. Harley E. H. (1995) DAPSA: DNA and Protein (2001) Phylogeny and historical biogeography of Sequence Analysis, Version 3.8. University of Lauraceae: evidence from the chloroplast and Cape Town, Cape Town. nuclear genomes. Ann. Missouri Bot. Gard. 88: Hooker J. D. (1890) The Flora of British , 104–134. Vol. 5. L. Reeve, London. Chase M. W., Cox A. V., Soltis D. E., Soltis P. S., Huelsenbeck J. P., Bull J. J., Cunningham C. W. Moil M. E., Savolainen V., Reeves G., Hoot S. B., (1996) Combining data in phylogenetic analysis. Morton C. M. (1997) Large DNA sequence TREE 11: 152–158. matrices, phylogenetic signal and feasibility: an Hutchinson J. (1964) The genera of flowering empirical approach. Amer. J. Bot., Suppl. 84: 181. plants, vol. 1. Clarendon Press, Oxford. Christophel D. C. (1994) The early Tertiary mac- Hyland B. P. M. (1989) A revision of Lauraceae in rofloras of continental Australia. In: Hill R. S. Australia (excluding Cassytha). Austral. Syst. (ed.) History of the Australian Vegetation: Bot. 2: 135–367. Cretaceous to Recent. Cambridge University Johnson L. A., Schultz J. L., Soltis D. E., Soltis P. Press, Cambridge, pp. 262–275. S. (1996) Monophyly and generic relationships J. Li et al.: Molecular phylogeny of the Litsea complex 33

of Polemoniaceae based on matK sequences. Meissner C. (1864) Lauraceae. In: de Candolle A. Amer. J. Bot. 83: 1207–1224. P. (ed.) Prodromus Systematis Naturalis Regni Johnson L. A., Soltis D. E. (1994) matK DNA Vegetablis, Vol. XV. Masson et Fils, Paris, sequences and phylogenetic reconstruction in pp. 1–260. Saxifragaceae s.str. Syst. Bot. 19: 143–156. Nees von Esenbeck C. G. D. (1836) Systema Johnson L. A., Soltis D. E. (1995) phylogenetic Laurinarum. Sumptibus Veitii et Sociorum, inference in Saxifragaceae sensu stricto and Gilia Berlin. (Polemoniaceae) using matK sequences. Ann. Nixon K. C., Carpenter J. M. (1996) On simulta- Missouri Bot. Gard. 82: 149–175. neous analysis. Cladistics 12: 221–241. Judd W. S., Campbell C. S., Kellogg E. A., Stevens Olmstead R. G., Palmer J. D. (1994) Chloroplast P. F. (1999) Plant systematics: a phylogenetic DNA systematics: a review of methods and data approach. Sinauer, Sunderland Mass. analysis. Amer. J. Bot. 81: 1205–1224. Kostermans A. J. G. H. (1957) Lauraceae. Pen- Pax F. (1889) Lauraceae. In: Engler A., Prantl K. gumuman Balai Besar Penjelidikan Kehutanan (eds.) Die natu¨rlichen Pflanzenfamilien Bd. III, Indonesia 57: 1–64. 2. Engelmann, Leipzig, pp. 106–126. Kron A. K. (1997) Phylogenetic relationships of Plunkett G. H., Soltis D. E., Soltis P. S. (1996) Rhododendroideae (Ericaceae). Amer. J. Bot. Evolution patterns in Apiaceae: inferences based 84: 973–980. on matK sequence data. Syst. Bot. 21: 477–495. Li H.-W. (1985) Parallel evolution in Litsea and Plunkett G. H., Soltis D. E., Soltis P. S. (1997) Lindera of Lauraceae. Acta Bot. Yunnan. 7: Clarification of the relationship between Apia- 129–135. ceae and Araliaceae based on matK and rbcL Li H.-W. (1995) The origin and evolution of Litsea sequence data. Amer. J. Bot. 84: 565–580. genera group (Laureae) in Lauraceae. Acta Bot. Rehder A. (1920) The American and Asiatic species Yunnan. 17: 251–254. of Sassafras. J. Arnold Arbor. 1: 242–245. Li H.-W., Pai P. Y., Lee S. K., Wei F. N., Wei Richter H. G. (1981) Anatomie des sekunda¨ren Y. T., Yang Y. C., Huang P. H., Tsui H. P., Shia Xylems und der Rinde der Lauraceae. Sonderb. Z. D., Li J. L. (1984) Lauraceae. In: Li H.-W. Naturwiss. Vereins Hamburg 5: 1–148. (ed.) Flora of Reipublicae Popularis Sinicae, Rohwer J. G. (1993) Lauraceae. In: Kubitzki K., Vol. 31. Science Press, Beijing. Rohwer J. G., Bittrich V. (eds.) The families and Li J. (2001) Systematic relationships Within the genera of vascular plants, Vol. 2. Springer, Litsea complex (Lauraceae). Ph.D. The Univer- Berlin, pp. 366–391. sity of Adelaide (unpublished). Rohwer J. G. (2000) Toward a phylogenetic Li J., Christophel D. C. (2000) Systematic relation- classification of the Lauraceae: evidence from ships within the Litsea complex (Lauraceae): a matK sequences. Syst. Bot. 25: 60–71. cladisitic analysis based on morphological and Rohwer J. G., Richter H. G., van der Werff H. leaf cuticle data. Austral. Syst. Bot. 13: 1–13. (1991) Two new genera of Neotropical Laura- Liang H. P., Hilu K. W. (1996) Application of the ceae and critical remarks on their generic delim- matK gene sequences to grass systematics. Ca- itation. Ann. Missouri Bot. Gard. 78: 388–400. nad. J. Bot. 74: 125–134. Steele K. P., Vilgalys R. (1994) Phylogenetic Long D. G. (1984) Notes relating to the Flora of analysis of Polemoniaceae using nucleotide Bhutan: VIII. Notes Roy. Bot. Gard., Edin- sequences of plastid gene matK. Syst. Bot. 19: burgh 41: 505–525. 126–142. Manos P. S., Steele K. P. (1997) Phylogenetic Swofford D. L. (1998) PAUP* Phylogenetic Anal- analyses of ‘‘higher’’ Hamamelididae based on ysis Using Parsimony. Sinauer and Associates, plastid sequence data. Amer. J. Bot. 84: 1407– Sunderland, Massachusetts. 1419. Tsui H. P. (1987) A study on the system of Lindera. Mason-Gamer R. J., Kellogg E. A. (1996) Testing Acta Phytotax. Sinica 25: 161–171. for phylogenetic conflict among molecular data van der Werff H. (2001) An annotated key to the sets in the tribe Triticeae (Gramineae). Syst. genera of Lauraceae in the Flora Malesiana Biol. 45: 524–545. region. Blumea 46: 125–140. 34 J. Li et al.: Molecular phylogeny of the Litsea complex van der Werff H., Richter H. G. (1996) Toward an Addresses of the authors: Jie Li, Hsi-Wen Li, improved classification of Lauraceae. Ann. Mis- Laboratory of Plant Phylogenetics and Conserva- souri Bot. Gard. 83: 409–418. tion Biology, Xishuangbanna Tropical Botanical Weberling F. (1985) Infloreszenzmorphologie der Garden, The Chinese Academy of Sciences, 88 Lauraceae. Bot. Jahrb. Syst. 107: 395–414. Xuefu Rd., Kunming, Yunnan 650223, P.R. China. White T. J., Bruns T. D., Lee S. B., Taylor J. W. D. C. Christophel, Department of Biological Sci- (1990) Amplification and direct sequencing of ence, University of Denver, 21909 E, Iliff, Denver, ribosomal RNA genes and the internal tran- CO 80208, USA. J. G. Conran (e-mail: john.con- scribed spacer in fungi. In: Innis M. A., Gelfand [email protected]), Centre for Evolutionary G. H., Sninsky J. J., White T. J. (eds.) PCR – Biology and Biodiversity, Discipline of Environ- Protocols and applications – a laboratory man- mental Biology, School of Earth and Environmen- ual. Academic Press, Orlando Fl., pp. 315–322. tal Sciences, Darling Building DP418. The Xiang Q. Y., Soltis D. E., Soltis P. S. (1998) University of Adelaide, SA 5005, Australia. Phylogenetic relationships of the Cornaceae and close relatives inferred from matK and rbcL sequences. Amer. J. Bot. 85: 285–297.