| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Systematics |
2State Key Laboratory, School of Life Sciences, Zhongshan University, Guangzhou 510275, People's Republic of China; 3Department of Botany, California Academy of Sciences, 875 Howard Street, San Francisco, California 94103 USA; 4The New York Botanical Garden, Cullman Program for Molecular Systematics Studies, 200th St. & Southern Blvd., Bronx, New York 10458 USA
Received for publication August 26, 2003. Accepted for publication August 3, 2004.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
300 species of woody flowering plants with a disjunct distribution between the warm-temperate to tropical regions of eastern Asia and the Americas. Phylogenetic analyses of 111 species of Symplocos based on the nuclear ribosomal internal transcribed spacer (ITS) region and the chloroplast genes rpl16, matK, and trnL–trnF yielded topologies in which only one of the four traditionally recognized subgenera (Epigenia; Neotropics) is monophyletic. Section Cordyloblaste (subgenus Symplocos; eastern Asia) is monophyletic and sister to a group comprising all other samples of Symplocos. Section Palura (subgenus Hopea; eastern Asia) is sister to a group comprising all other samples of Symplocos except those of section Cordyloblaste. Symplocos wikstroemiifolia (eastern Asia) and S. tinctoria (southeastern United States), both of subgenus Hopea, form a clade that groups with S. longipes (tropical North America) and the species of subgenus Epigenia. The remaining samples of subgenus Hopea (eastern Asia) form a clade. Section Neosymplocos (subgenus Microsymplocos; Neotropics) is well nested within a clade otherwise comprising the samples of section Symplocastrum (subgenus Symplocos; Neotropics). Section Urbaniocharis (subgenus Microsymplocos; Antilles) groups as sister to the clade comprising Symplocastrum and Neosymplocos. The data support the independent evolution of deciduousness among section Palura and S. tinctoria. The early initial divergence of sections Cordyloblaste and Palura from the main group warrants their recognition at taxonomic levels higher than those at which they are currently placed. An inferred eastern Asian origin for Symplocos with subsequent dispersal to the Americas is consistent with patterns from other phylogenetic studies of eastern Asian-American disjunct plant groups but contrary to a North American origin inferred from the earliest fossil occurrences of the genus.
Key Words: disjunction • ITS • matK • phylogeny • rpl16 • Symplocaceae • Symplocos • trnL–trnF
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
300 species of woody flowering plants distributed in the New World and the lands bordering the western Pacific Rim (Brand, 1901
; Nooteboom, 1975
; Ståhl, 1995
). It is found primarily in humid tropical montane forests. Two taxa reach the north-temperate zone, one in eastern Asia (section Palura of subgenus Hopea, to 45° N) and one in the southeastern United States (S. tinctoria, to 37° N). These are the only taxa with the deciduous condition in the genus. In the Old World, Symplocos occurs no farther west than India. Its distribution corresponds to an "amphi-Pacific tropical" pattern of disjunction documented in 100 genera and higher groups of seed plants (Steenis, 1962
, 1963
; Thorne, 1972
).
Symplocos is recognized as the sole genus of Symplocaceae by recent authors (Cronquist, 1981
; Takhtajan, 1997
; Thorne, 2000
), although the family has been divided into many genera in the older literature (e.g., Miers, 1880
; Nakai, 1922
, 1927
; Hatusima, 1936
). One of 24 families comprising order Ericales sensu the Angiosperm Phylogeny Group (1998
, 2003
), the Symplocaceae are characterized by leaves without stipules; sympetalous, actinomorphic flowers; epipetalous, united, and numerous stamens; an inferior, incompletely loculed ovary; a simple style; unitegmic ovules; and a drupaceous fruit with a hard endocarp (Nooteboom, 1975
). No other group of Ericales possesses this combination of characters, leaving little doubt that the family is monophyletic.
In the most recent comprehensive taxonomic revision of Symplocos, Brand (1901)
recognized 277 species within four subgenera: Symplocos, Epigenia, Hopea, and Microsymplocos. In subgenus Symplocos (76 species, eastern Asian and American tropics), the androecium is adnate to the corolla for approximately half its total length, the stamens are monadelphous, and the filaments are complanate. In the other subgenera, the androecium is adnate to the corolla only at the base. In subgenus Epigenia (20 species, Neotropics), the stamens are partly to completely distinct and the filaments are filamentous. In subgenus Hopea (167 species, eastern Asia except S. tinctoria), the stamens are pentadelphous and the filaments are filamentous. In subgenus Microsymplocos (14 species, Neotropics), the stamens are monadelphous, the filaments are claviform, and the flowers are uniformly small (2 to 5 mm long).
Brand (1901)
divided each subgenus into two sections. Within subgenus Symplocos, section Cordyloblaste (six species, tropical eastern Asia) is distinguished from section Symplocastrum (70 species, Neotropics) by the filaments completely connate or distinct only for a short distance distally (vs. distally distinct from the point of separation from the corolla), and the ovary is two-locular (vs. three- to five-locular). Within subgenus Epigenia, section Barberina (17 species, southern Brazil) differs from section Pseudosymplocos (three species, Antilles) by the presence of morphological androdioecy (vs. hermaphroditism), up to 25 (vs. more numerous) stamens per flower, and the cotyledons shorter than the radicle (vs. longer). Within subgenus Hopea, section Palaeosymplocos (six species, eastern Asia and eastern North America) differs from section Bobu (161 species, eastern Asia) by its distinctly (vs. indistinctly) pentadelphous stamens. Within subgenus Microsymplocos, section Neosymplocos (11 species, southern Brazil) differs from section Urbaniocharis (three species, Antilles) by its pubescent (vs. glabrous) filaments and four-locular (vs. two-locular) ovary.
Several authors of regional treatments of Symplocos produced since the time of Brand's (1901)
monograph have revised the infrageneric classification. In revisions of Symplocos of the Old World, Nooteboom (1975
, 1977
, 1980
) pared the number of Asian species to 116 from Brand's 172 (including a reduction of section Cordyloblaste to two species) and revised Brand's subgeneric classification. Nooteboom (1975)
placed subgenera Epigenia and Microsymplocos under subgenus Hopea based on palynological and scant phytochemical data. In a revision of the Japanese species of Symplocos, Nagamasu (1989a
, b
, 1993
) recognized subgenus Microsymplocos and placed subgenus Epigenia under subgenus Hopea based on palynological characters. Bidá (1995)
, in revising the Brazilian species of the genus, upheld Brand's (1901)
classification, citing unspecified floral and fruit characters and the palynological work of Barth (1979)
as justification. These works represent significant advances in understanding the systematics of Symplocos but are limited by a focus only on the subgenera occurring within the geographic region under consideration.
Of Brand's (1901)
sections, only those in subgenus Hopea have been disputed. Handel-Mazzetti and Peter-Stibal (1943)
recognized sections Bobu, Lodhra, Palaeosymplocos, and Palura in a treatment of the Chinese species of Symplocos based on the following characters: leaf midrib adaxially impressed (vs. prominent); fruit shape; disk glabrous (vs. pubescent); and ovary two- (vs. three-) locular. Wu (1986a
, b
, 1987
) added sections Glomeratae and Singuliflorae to the system of Handel-Mazzetti and Peter-Stibal (1943)
based on a glomerulate and one-flowered inflorescence, respectively. Based on various morphological characters, Nagamasu (1993)
classified the Japanese species of subgenus Hopea into eight sections, of which three (Glaucae, Lancifoliae, and Okinawenses) were newly described.
A phylogenetic study of 29 species of eastern Asian Symplocos with DNA sequence data from the nuclear ribosomal internal transcribed spacer (ITS) region, and the chloroplast trnL–trnF region (comprising the trnL [UAA] intron, the trnL [UAA] 3' exon, and trnL [UAA]–trnF [GAA] spacer) and trnH–psbA spacer (Soejima and Nagamasu, 2004
) has provided valuable insight into the classification and evolution of Symplocos in Japan. Because subdivisional sampling was limited, however, this study could not address the global classification of the genus. Here we test the various subdivisional classifications of Symplocos by including representatives from all subgenera and described sections of Symplocos in a phylogenetic analysis of the genus. We use DNA sequence data from the ITS region and three regions of the chloroplast genome: the rpl16 intron, the matK gene, and the trnL–trnF region. Because of their relative ease of use and comparatively rapid sequence evolution, these regions have become standard sources of nucleotide characters in lower-level phylogenetic studies of plants. From the resultant phylogenetic patterns we explore character evolution in Symplocos (particularly that of leaf persistence, carpel number, and pedicel articulation) and the historical biogeography of the genus in the context of its amphi-Pacific disjunct distribution.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
except for the sectional classification of the Chinese and Japanese species of subgenus Hopea, which follows Wu (1987)
and Nagamasu (1993
; Appendix, see Supplemental Data with the online version of this article). DNA sequences from 120 accessions of Symplocos were newly generated for this study (Appendix). This sample includes representatives of 90 Symplocos species, all four subgenera sensu Brand, and all sections recognized by Brand (1901)
, Wu (1987)
, and (in combination with the ITS data of Soejima and Nagamasu [2004]
) Nagamasu (1993)
.
The sister group of Symplocaceae has long remained uncertain (Anderberg et al., 2002
) but a recent molecular study of Ericales supports a clade consisting of Symplocaceae, Styracaceae, and Diapensiaceae and a more-inclusive clade comprising Theaceae, Roridulaceae, Actinidiaceae, Sarraceniaceae, Clethraceae, Cyrillaceae, and Ericaceae (Schöneberger et al., 2004
). We chose the genus Ternstroemia (Theaceae [Cronquist, 1981
] or Ternstroemiaceae [Angiosperm Phylogeny Group, 1998
, 2003
]) as outgroup for the combined chloroplast DNA and combined four-gene analyses because it is one of the few genera within Ericales for which sequences of all three chloroplast genes are available. Moreover, the genetic distance between Symplocos and Ternstroemia is low relative to those between Symplocos and representatives of 10 other families of Ericales (see Fritsch et al., 2001
). The ITS analysis included 31 samples of Symplocos generated from Soejima and Nagamasu (2004)
. To further confirm the root of the Symplocos tree, we conducted phylogenetic analyses on separate ITS and matK data sets that included representatives of various families of Ericales as outgroups (Appendix).
Total DNAs were isolated with DNeasy Plant Mini DNA extraction kits (Qiagen, Inc., Valencia, California, USA) or with the cetyltrimethyl ammonium bromide (CTAB) method of Doyle and Doyle (1987)
from desiccant- or air-dried leaf tissue. Some leaf samples (20 mg) were obtained from herbarium specimens (Appendix). Prior to extraction, dried leaf tissue was pulverized by high-speed action of the Wig-L-Bug grinding mill (REFLEX Analytical Corp., Ridgewood, New Jersey, USA). PCR amplification was performed with standard methods (Dieffenbach and Dveksler, 1995
) and BIOLASE (Bioline USA, Inc., Randolph, Massachusetts, USA) as the DNA polymerase. PCR products were purified with the Wizard PCR Preps DNA purification system (Promega, Madison, Wisconsin, USA). Cycle sequencing was performed with the ABI Prism BigDye Terminators v2.0 cycle sequencing reaction kit (Applied Biosystems, Foster City, California, USA) by using 1/4-scale reaction mixtures in a model 9600 PCR system thermal cycler (Perkin-Elmer, Boston, Massachusetts, USA). Sequences were determined with an ABI Prism 3100 genetic analyzer (Applied Biosystems) by obtaining forward and reverse reads for all samples. Sequences were edited with the computer program Sequencher (3.0 and 4.1; Gene Codes, Ann Arbor, Michigan, USA) and all sequences have been deposited in GenBank (Appendix).
Amplification and sequencing of the ITS region employed primers ITS–4, ITS–5p, ITS–2p, and ITS–3p from Swensen et al. (1998)
. The rpl16 intron was amplified with the forward primer F71 (5'–GCTATGCTTAGTGTGTGACTCGTTG–3'; Jordon et al., 1996
) and the reverse primer L16 exon2 (Downie et al., 2000
). Some samples were amplified with the forward primer rps3 (Downie et al., 2000
) instead of F71. rpl16 sequencing was performed with F71 and L16exon2, and the following internal primers designed specifically for Symplocos: 221F (5'–CTGATTCTAAGTTGTGAAGC–3'), 618F (5'–GCCGGGAAGCAATTAATCTA–3'), 191R (5'–TATTTCAGTTGTTACAATTA–3'), and 609R (5'–CCATCCCGACCAATGAATCA–3'). The matK gene was amplified with the primers matK–1F and matK–1R (Sang et al., 1997
). The amplified region includes the entire matK coding region and a total of 300 bp of the trnK intron that flanks the ends of matK (70 bp at the 5' end and 230 bp at the 3' end). matK sequencing was performed with the two amplification primers and the internal primers symp–1176F (5'– CAATTCATTCAMTATTTCCTT–3'), symp–1540F (5'–GTTCAAGGATCCTTTCATGC–3'), symp–2030F (5'–CTTCGACTTTCTTGTGCTAG– 3'), symp–1988R (5'–ACGCCCGAATCGGTCAATAA–3'), symp–1470R (5'–AAGATATTAATCGTAAATGA–3'), and symp–866R (5'–CTATGATCATGAGCAAGTGC–3') designed specifically for Symplocos. The trnL–trnF region was amplified and sequenced with the universal primers c, d, e, and f (Taberlet et al., 1991
). Target sequences unsuccessfully amplified with the external primers were often successfully amplified in two fragments with an external and one of the internal primers.
Data analysis
Sequence alignment was manual except in the ITS analysis. The computer program ClustalX (Thompson et al., 1997
) was used for ITS alignment because high sequence divergence among the outgroups and between the outgroups and ingroup made manual alignment problematic. Gap opening and gap extension parameters were 10 and 0.2, respectively. Aligned sequence matrices are available from the authors upon request. The computer program MacClade version 4.0 (Maddison and Maddison, 2000
) was used to translate DNA sequences into protein sequences to aid the alignment of matK sequences. Gaps introduced into the alignment were treated as missing data. Various preliminary analyses with unambiguously aligned parsimony-informative gaps included had little effect on topology and clade support; these characters were therefore not included in final analyses.
Phylogenetic analyses employed maximum parsimony (MP) and Bayesian inference (BI). The MP analyses were conducted with the heuristic search option in PAUP* version 4.0b10 (Swofford, 2002
). Searches were conducted over 100 random-taxon-addition replicates with tree bisection-reconnection branch-swapping, steepest descent, and MulTrees in effect. All characters and states were weighted equally and unordered. All trees from the replicates were swapped to completion, all shortest trees were saved, and a strict consensus tree was computed. To provide a complete search for the shortest tree in the ITS analysis, MulTrees was not enforced and no more than 3000 trees were saved per replicate. Relative support for individual clades was estimated with the parsimony bootstrap (bt) method (Felsenstein, 1985
). One thousand pseudoreplicates were performed with uninformative characters excluded. Ten random-taxon-addition heuristic searches for each pseudoreplicate were performed and all minimum-length trees were saved per search. To expedite the search, MulTrees was not enforced in the ITS bootstrap analysis.
Bayesian analyses were conducted with MrBayes version 3.0b4 (Huelsenbeck and Ronquist, 2001
) by using uniform prior probabilities and estimating base frequencies and the parameters for the HKY + 
model. We ran four chains of the Markov chain Monte Carlo by beginning with a random tree and sampling one tree every 100 generations for 1 000 000 generations. The first 30 000 generations of the chain were used as "burn in" after stationarity was reached, and the phylogenetic estimate was based on trees sampled after generation 30 000. To estimate the posterior probability (pP) of recovered branches, 50% majority-rule consensus trees were created. An ITS phylogram was produced as average-branch-length consensus trees with MrBayes.
Data set congruence was determined with incongruence length difference (ILD) tests (Farris et al., 1994
, 1995
; implemented in PAUP* as the partition-homogeneity test) by using 500 heuristic-search randomizations, simple addition sequence, and TBR swapping, with uninformative characters excluded, one tree held at each step, MulTrees not enforced, and no more than 2000 trees saved per randomization. We conducted pairwise comparisons and a three-way comparison with the chloroplast data sets. We also conducted a pairwise comparison with the ITS data set and the combined chloroplast data set, and a four-way comparison of all data sets. After assessing data congruence, data sets were combined to provide a total-evidence phylogenetic estimate. Samples missing all data in one or more of the individual data sets (see Appendix) were excluded from combined analyses except in some analyses involving Symplocos longipes and S. tenuifolia.
We used Fitch parsimony optimization (Maddison et al., 1992
) to assess the historical biogeography of Symplocos with the molecular topology. This method assumes that geographic distributions are solely the result of dispersal (as opposed to vicariance) events. Thus, polymorphic area states are restricted to terminal nodes. Ancestral states are inferred through minimizing the number of character state changes on the tree. The data matrix was constructed by coding "area" as a single multistate character, and the analysis was performed with MacClade 4.0 (Maddison and Maddison, 2000
). The four areas circumscribed for the analysis were eastern Asia, North America, South America, and the Antilles. The combined four-gene tree was used for optimization, with all clades comprising operational taxonomic units (OTUs) from the same area reduced to individual OTUs. This reduction will have no effect on the interpretation of the results in the context of our biogeographic objectives. Analyses were conducted with S. longipes in either of two possible artificially resolved positions.
The evolution of ovary cell number and leaf persistence was inferred with Fitch parsimony optimization onto the combined four-gene reduced topology. Character states were scored for each OTU and the characters were optimized with MacClade 4.0 (Maddison and Maddison, 2000
).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
ITS analysis
The ITS data set included 143 OTUs, including 31 samples of Symplocos from Soejima and Nagamasu (2004)
and eight outgroups, representing a total of 109 Symplocos species. The ITS region is the most length-variable of all gene regions analyzed in this study. Total sequence length in Symplocos ranges from 609 base pairs (bp; e.g., S. celastrinea) to 635 bp (e.g., S. lanata). The length of ITS 1 ranges from 246 bp (e.g., S. berteroi) to 253 bp (S. chinensis and S. paniculata). The 5.8S gene is consistently 164 bp long. The length of ITS 2 in Symplocos ranges from 198 bp (e.g., S. celastrinea) to 220 bp (e.g., S. confusa). All samples of subgenus Epigenia have a 20-bp deletion in ITS 2. Alignment with all outgroups resulted in a data set of 694 characters, of which 447 (64.4%) were variable and 339 (48.8%) were parsimony-informative.
Symplocos is monophyletic in both MP (1700 equally parsimonious trees of length = 1409, consistency index [CI] = 0.45, retention index [RI] = 0.78; bt = 100) and BI analyses of ITS sequences (pP = 0.93; Fig. 1A). The species of section Cordyloblaste sampled form a clade (pP = 0.96; bt = 100; Fig. 1B) that is sister to the clade (pP = 0.93; bt = 100) comprising all other Symplocos species. Other well-supported major clades comprise (1) section Palura (pP = 0.95; bt = 100); (2) the species of subgenus Epigenia (pP = 0.94; bt = 100); (3) the species of section Barberina (pP = 0.94; bt = 88); (4) the species of section Symplocastrum and Urbaniocharis (pP = 0.95; bt = 78); (5) most Asian species of subgenus Hopea ("main Asian Hopea" clade; pP = 0.94; bt = 81; Fig. 1B, 1C); and (6) S. tinctoria and S. wikstroemiifolia (pP = 0.96; bt = 68; Fig. 1B). Other than the first divergence, the basal nodes of the Symplocos ITS topology are not well supported.
|
The combined chloroplast data set (matK, rpl16, and trnL– trnF) included 72 OTUs. The length of the rpl16 intron in Symplocos ranges from 1008 bp in S. confusa to 1057 bp in S. celastrifolia. The aligned sequences comprised 1095 nucleotide characters (excluding the first 13 bp, which were often not readable or constant in those that were), of which 138 (12.6%) were variable and 49 (4.5%) were parsimony-informative. The length of the matK amplified fragment in Symplocos ranges from 1810 bp (e.g., S. wikstroemiifolia; including a coding region of 1506 bp) to 1831 bp (S. confusa; including a coding region of 1527 bp). The aligned sequences comprised 1855 nucleotide characters, of which 268 (14.4%) were variable and 78 (4.2%) were parsimony-informative. The length of the trnL–trnF region ranges from 905 bp in S. paniculata to 939 bp in S. quitensis. The length of the trnL intron ranges from 484 bp in S. paniculata to 518 bp in S. quitensis; that of the trnL [UAA] 3' exon is consistently 50 bp; and that of the trnL–trnF spacer ranges from 354 bp in, e.g., S. tortuosa to 365 bp in S. macrophylla. The aligned sequences comprised 974 nucleotide characters, of which 115 were variable (11.8%) and 19 (2.0%) were parsimony-informative.
In both the MP (1899 equally parsimonious trees of length = 234, CI = 0.68, RI = 0.89) and BI analyses of combined chloroplast sequences, Symplocos confusa, representing section Cordyloblaste, is sister to the clade (pP = 0.99; bt = 100) comprising all other Symplocos species (Fig. 2). Other well-supported major clades comprise (1) section Palura (pP = 1.00; bt = 100); (2) the sister clade to section Palura (pP = 98; bt = 85); (3) the species of subgenus Epigenia (pP = 0.97; bt = 92); (4) the species of section Barberina (pP = 0.97; bt = 98); (5) subgenus Epigenia, S. tinctoria, and S. wikstroemiifolia ("WTLE" clade; pP = 0.97; bt = 94); (6) section Urbaniocharis (pP = 0.99; bt = 71); and (7) the main Asian Hopea clade (pP = 0.99; bt = 99).
|
The combined four-gene data set included 70 OTUs. The aligned sequences comprised 4594 nucleotide characters, of which 777 (17.0%) were variable and 303 (6.6%) were parsimony-informative. Both the MP (128 equally parsimonious trees of length = 778, CI = 0.61, RI = 0.84) and BI analyses of the four-gene combined sequences yielded trees with higher clade resolution and support than those from ITS or chloroplast analyses alone (Fig. 3). Section Cordyloblaste is sister to the clade comprising all other Symplocos species (pP = 0.99; bt = 100). Within the latter, section Palura forms a clade (pP = 0.99; bt = 100) that is sister to the clade comprising the remaining species (pP = 0.99; bt = 94).
|
Analysis of Symplocos longipes and S. tenuifolia
Symplocos longipes and S. tenuifolia are significant for both classification and historical biogeography, because S. longipes has not been classified as to subdivision, and S. tenuifolia is the only representative of section Neosymplocos (subgenus Microsymplocos). The positions of S. longipes and S. tenuifolia sampled for a subset of the four genes (through lack of amplification for the missing gene regions) were assessed in separate analyses. In a combined analysis of the ITS and rpl16 data sets, S. longipes forms a trichotomy with the clade comprising subgenus Epigenia and that comprising S. tinctoria and S. wikstroemiifolia (Fig. 3; pP = 0.96, bt = 63 [not shown])
. In a combined analysis of the ITS and matK data sets, S. tenuifolia groups in a clade with all samples of section Symplocastrum (S. tenuifolia + section Symplocastrum, pP = 0.99; bt = 0.91). In an additional parsimony analysis, S. longipes and S. tenuifolia were included with all OTUs having complete four-gene data. This analysis yielded a topology with S. longipes in the same position as in the combined ITS and rpl16 analysis (bt = 84) and S. tenuifolia as highly nested within the rest of South American subgenus Symplocastrum as sister to S. fuscata (Fig. 3; bt = 89 [not shown]).
Biogeographical analysis
Fitch parsimony optimization with Symplocos longipes resolved as sister to subgenus Epigenia resulted in three equally optimal reconstructions of seven steps (Fig. 4A). The three reconstructions differ only in the assignment of the stem lineage of the clade comprising sections Pseudosymplocos and Barberina (Antilles, North America, or South America).
|
Fitch parsimony optimization with Symplocos longipes resolved as sister to the clade comprising S. tinctoria and S. wikstroemiifolia resulted in four equally optimal reconstructions of seven steps (not shown). Three of the four optimizations are identical to those reconstructed from the analysis with S. longipes resolved as sister to subgenus Epigenia. In the fourth, a dispersal from eastern Asia to the Antilles is inferred along the stem lineage immediately above the divergence of the main Asian Hopea clade. This is followed by two dispersals to North America, one along the stem of the clade comprising S. longipes, S. tinctoria, and S. wikstroemiifolia, the other along the stem immediately above the divergence of section Urbaniocharis. The remaining dispersals are identical to those inferred from other optimization analyses.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
; Hoot et al., 1999
), the four-gene parsimony analysis of Symplocos had a shorter computer run time and resulted in more highly resolved trees and higher support values than in analyses with separate data sets, although we caution that the higher number of taxa in some of the data sets vs. the four-gene combined data set could have also affected run times.
Subgeneric classification and character evolution
The combined analyses support only one of the four subgenera of Symplocos (subgenus Epigenia) recognized by Brand (1901)
. The data support neither subgenus Epigenia (Nooteboom, 1975
) nor subgenus Microsymplocos (Nooteboom, 1975
; Nagamasu, 1989a
, b
, 1993
) as synonyms of subgenus Hopea. The characters on which previous classifications have been based are assessed below in the context of their use in classification and their evolution. This is done with the understanding that the best way to resolve optimizations of morphological characters is to incorporate a morphological data matrix into studies of Symplocos, the construction of which is in progress (L. M. Kelly et al., unpublished data).
Subgenus Symplocos
The widely separated placement of sections Cordyloblaste and Symplocastrum in the topologies suggests that some or all of the characters used to define subgenus Symplocos are homoplasious or plesiomorphic. These characters include the androecium adnate to the corolla for approximately half its total length (vs. only at the base), stamens monadelphous (vs. pentadelphous), stamen filaments complanate (vs. filamentous), and (a character mentioned in Nooteboom [1975]
) stamen filaments constricted (vs. gradually tapered) distally. Upon further investigation, we have observed several characters that appear to distinguish section Cordyloblaste from all other species of the genus: calyx teeth apices truncate (vs. rounded to acute), corolla lobes papillate (vs. smooth) adaxially, ovary semi-inferior (vs. completely inferior), and an articulation between the hypanthium and the pedicel absent (vs. present). Whereas the first three characters may be autapomorphic for the section, the last is probably plesiomorphic in Symplocos because articulated pedicels are both uncommon and scattered throughout the Ericales clade, and nonarticulated pedicels therefore are likely to be the plesiomorphic state of the entire order (Fritsch et al., 2001
). As such, the presence of articulation is a synapomorphy for the non-Cordyloblaste clade in Symplocos.
Chromosome and pollen data further distinguish section Cordyloblaste from the rest of the genus. The chromosome number of section Cordyloblaste is 2n = 90 vs. 2n = 22, 24, and possibly 28 for the rest of the species of the family that have been sampled (Nooteboom, 1975
; Nagamasu, 1993
), suggesting that this section is octoploid. No taxa of section Symplocastrum have yet been sampled cytologically for comparison. The three-colporate pollen grains of section Cordyloblaste with a massive tectum and a smooth perforate exine (Henschelii subtype; Meijden, 1970
; Liang, 1986
) differ from those of section Symplocastrum (type 2: two-colporate, exine smooth or more or less punctate; and type 4: three-colporate, exine verrucose; Barth, 1979
, 1982
). They also differ from those of subgenus Epigenia (three-colporate, tectum reduced, exine baculate and finely or coarsely verrucose) and subgenus Hopea (tectum thin, exine with supratectal structure; Meijden, 1970
; Liang, 1986
).
Section Cordyloblaste has not been accorded generic status since its initial treatment at the generic level (Moritizi, 1848
; Miers, 1880
). Nonetheless, the early diversification of this group in combination with its unique morphological features and chromosome number provide justification for reassessing the level at which this taxon is best recognized. Variation in pedicel articulation may prove to be particularly informative in this regard. The presence of articulated pedicels is a synapomorphy of a major clade above the genus level within Styracaceae (Fritsch et al., 2001
) and appears to be genus-specific throughout most of the order Ericales (P. W. Fritsch, unpublished data). The recognition of Cordyloblaste at the genus level is therefore generally consistent with overall generic concepts in Ericales.
Subgenus Hopea
Section Palura of subgenus Hopea sensu Wu (1987)
is treated by various authors as constituting anywhere from one to five species (Nooteboom, 1975
; Wu, 1987
; Wu and Nooteboom, 1996
; Nagamasu, 1993
). It is the only group in Symplocos that possesses exclusively terminal panicles and suprareticulate pollen exine (Barth, 1979
, 1982
; Liang, 1986
; Nagamasu, 1993
), corroborating its monophyly as supported here. The deep divergence of section Palura in the topology was not predicted from the classifications of Brand (1901)
, who placed the complex in section Bobu subsection Palura with many other species, or Wu (1987)
and Nagamasu (1993)
, who placed it in its own section within subgenus Hopea. Two morphological characters, however, corroborate our results. Section Palura has a chromosome number of n = 11, a probable synapomorphy for all Symplocos species excluding those of section Cordyloblaste. It also has a two-locular ovary (vs. three- to five-locular in the rest of the genus). Two-locular ovaries are shared among the members of sections Palura, Cordyloblaste, and Urbaniocharis, and some members of the main Asian Hopea clade (Brand, 1901
; Nagamasu, 1993
). Contrary to the hypothesis of Nagamasu (1993)
, our results unambiguously support the ancestral two-locular condition in Symplocos followed by various equally optimal scenarios that include a reversal to the two-locular condition (Fig. 4B).
Miers (1880)
grouped the Asian deciduous species (section Palura sensu Wu, 1987
) with Symplocos pendula of section Cordyloblaste sensu Brand (1901)
as the sole members of the genus Palura. Although our data do not support the genus Palura sensu Miers (1880)
, the deep divergence of the Asian deciduous clade within Symplocos does support Miers' recognition of this group at a level higher than section. If section Cordyloblaste were raised to the generic level, then it would be reasonable to recognize section Palura sensu Wu (1987)
as one of two subgenera of Symplocos, the other comprising the members of its sister group.
Section Palura is one of only two deciduous elements in Symplocos, the other being S. tinctoria (also in subgenus Hopea) of the southeastern United States. Symplocos tinctoria is deciduous in the northern portion of its range. In warmer areas to the south, the leaves from the previous season remain through the spring flowering period and the unfolding of new leaves (Elias, 1980
); thus, S. tinctoria in these regions is considered evergreen or nearly so (Peattie, 1950
; Radford et al., 1964
; Elias, 1980
; Little, 1980
). Section Palura appears to be consistently deciduous throughout its range. Because section Palura and S. tinctoria are placed in widely separated regions of the topology, the data support the independent evolution of deciduousness from the evergreen condition in these lineages (Fig. 4C).
The WTLE clade does not accord with classifications of Symplocos. Some aspects of pollen morphology, however, appear to support this clade. The three-colporate pollen type with verrucose tectum occurs in both S. wikstroemiifolia and subgenus Epigenia (Barth, 1982
; Liang, 1986). The pollen of S. tinctoria is also three-colporate but has a thin perforate tectum, distinct supratectal ornamentation, and a thin but distinct columella layer (Nagamasu, 1989a
). This wall structure is the same as Barth's type 2 celastrinea type characterizing subgenus Epigenia (Nagamasu, 1989a
), suggesting a close relationship between S. tinctoria and subgenus Epigenia. The pollen of S. longipes has not been described. The combined data place the WTLE clade as sister to the clade comprising section Symplocastrum and subgenus Microsymplocos. Clade support for this placement, however, is inconclusive, with a high posterior clade probability (0.98) but a fairly low bootstrap percentage (64; Fig. 3). The alternative position of the WTLE clade as sister to the main Asian Hopea clade would accord better with the placement of subgenus Epigenia as a synonym of subgenus Hopea, as in Nooteboom (1975)
and Nagamasu (1993)
. Higher confidence regarding the placement of this clade will require more phylogenetic data.
The sister-group relationship between Symplocos tinctoria and S. wikstroemiifolia (continental southeastern Asia, including southern China) conflicts with previous research on the systematics of Symplocos. Brand (1901)
and Nooteboom (1975)
suggested an affinity of S. tinctoria with S. lucida, a complex treated as one (Nooteboom, 1975
; Wu and Nooteboom, 1996
) or seven (Wu, 1987
) species. Wu (1987)
placed S. wikstroemiifolia in the same section (Palaeosymplocos) as a member of the S. lucida complex but made no mention of S. tinctoria. Our data place the S. lucida complex within the main Asian Hopea clade instead of with S. tinctoria (Figs. 2, 3). Like molecular data from other groups with a distribution in eastern Asia and eastern North America (see Wen, 1999
), those from Symplocos appear to reject a disjunct species-pair hypothesis based on morphology in favor of an alternative set of relationships (in our case, a different species pair). Whether this is due to morphological stasis (symplesiomorphy), convergent evolution, or another process has yet to be assessed.
Neither Brand's (1901)
nor Wu's (1987)
sectional classifications are supported by the data. Of Wu's (1987)
seven sections in subgenus Hopea, only section Palura (see above) is unambiguously supported as monophyletic. Within section Palaeosymplocos, the data support the monophyly of the Symplocos lucida complex, but S. anomala groups as the first-diverging lineage within the main Asian Hopea clade (four-gene analysis), S. wikstroemiifolia groups in the WTLE clade (cpDNA and four-gene), and S. groffii is nested within a clade otherwise comprising species of sections Lodhra (S. nokoensis, S. sumuntia, and S. viridissima) and Singuliflorae (S. ovatilobata; cpDNA and four-gene). Symplocos euryoides, the other species in section Singuliflorae, groups strongly with the species of section Glomeratae and S. adenophylla of section Lodhra (four-gene analysis). The samples of section Bobu occur in three separate clades (all analyses). The samples of section Glomeratae form two clades that are unresolved with respect to a clade comprising S. adenophylla and S. euryoides in the highest-resolution analysis (four-gene analysis). Neither of Brand's (1901)
sections of subgenus Hopea are supported because section Palaeosymplocos includes S. tinctoria and section Bobu includes the Asian deciduous species (Wu's section Palura).
Because Nagamasu's (1993)
sections of subgenus Hopea were delimited only in the context of the Japanese species of Symplocos, they are difficult to assess with more-inclusive phylogenetic data. The three new sections described by Nagamasu each consist of one or two species. Section Okinawenses (containing only S. okinawensis) groups with S. anomala. Two samples of section Lancifoliae were analyzed (both S. lancifolia; the only other species in the section, S. microcalyx, was not sampled). The sample from Soejima and Nagamasu (2004)
groups with S. formosana (section Lodhra), whereas ours groups with S. mollifolia (section Bobu). Similarly, the two samples of section Glaucae (both S. glauca) occur in different positions. Because we have not seen vouchers from the study of Soejima and Nagamasu (2004)
, we cannot discern the reason for the different positions of the samples of these sections.
Although our data provide resolution of some of the relationships within the main Asian clade of subgenus Hopea (i.e., section Glomeratae, the Symplocos lucida complex, and several small clades), the clade is internally neither well-resolved nor strongly supported in any analysis. If clades within this group are to be formally recognized, parts of the topology suggest possible recircumscriptions. For example, section Glomeratae sensu Wu (1987)
may be expanded to accommodate all species with glomerulate inflorescences (e.g., S. adenophylla, S. adenopus, S. austrosinensis, S. congesta, S. euryoides, S. glandulifera, S. glauca, S. glomerata, S. grandis, S. poilanei, S. spectabilis, and S. stellaris, as sampled here). Also, the small sections Glaucae, Lancifoliae, and Okinawenses may be best treated as synonyms because they each group strongly with species from other sections. Although visual inspection of the four-gene data set suggests that homoplasy accounts for some of the lack of resolution, most clade ambiguity appears to be from low sequence divergence, suggesting rapid speciation in this clade relative to others that exhibit higher internal resolution. In this context, the data support Nooteboom's (1975)
view of avoiding sectional delimitation in subgenus Hopea altogether as the most appropriate taxonomic approach to the main Asian Hopea clade.
Subgenus Microsymplocos
The apparent polyphyly of subgenus Microsymplocos conflicts with the characters used by Brand (1901)
to define this subgenus. Observations on the species of section Urbaniocharis (P. W. Fritsch, unpublished data), however, contradict two of Brand's character descriptions for subgenus Microsymplocos. All species examined possess a pentadelphous rather than monadelphous arrangement of the stamens, and the filaments are filamentous, not claviform. Because more inclusive clades from the molecular data exclusively have these states as well (i.e., all except those of subgenus Symplocos), the placement of section Urbaniocharis as sister to section Symplocastrum supports the ancestral condition of these characters in section Urbaniocharis. From field and herbarium specimen observations, characters that appear to be unique to section Urbaniocharis (and possibly subgenus Microsymplocos) are the cupuliform corolla, stamens more or less appressed to the inner corolla surface, and the stamens shorter than the corolla, all of which support the monophyly of the section.
Field observations are currently lacking for species of section Neosymplocos. From examination of herbarium specimens, however, the filaments appear flattened and at least sometimes strongly constricted at the apex, as in section Symplocastrum. The nested placement of section Neosymplocos within section Symplocastrum supports the shared derived status of these characters throughout this clade. The data support the independent evolution of small flowers common to all species of sections Neosymplocos and Urbaniocharis. This convergence may have been influenced by the predominance of minute endemic insects that serve as pollinators for much of the flora of the Greater Antilles (Borhidi, 1996
).
The pollen of section Neosymplocos is similar to that of section Symplocastrum in its indistinct columellae and massive tectum without supratectal ornamentation (Barth, 1979
, 1982
), consistent with our results that group these two sections together. Also consistent with our data, Mai (1986)
has distinguished a "tenuifolia" type of pollen characterizing section Neosymplocos from the "celastrinea" and "lanata" types that are found in section Urbaniocharis. The tenuifolia type has a triangular-suboblate grain with reticulate-punctate exine, whereas those of the celastrinea and lanata types have a circular-oblate or -suboblate grain with verrucose or areolate exine, respectively (Mai, 1986
).
The only chromosome number known for subgenus Microsymplocos is n = 12 (Symplocos micrantha of section Urbaniocharis). This number differs from all non-Cordyloblaste Symplocos sampled (n = 11 and possibly n = 14). It is unclear whether the interpretation of the S. micrantha chromosome data is correct or, as Nooteboom (1975)
surmised, erroneous from the presence of B chromosomes.
Biogeographical implications
The earliest record of Symplocos known is fossil pollen from the western United States dating from the Late Cretaceous (Campanian-Maastrichtian, 84–65 million years ago), but there remain doubts as to whether this pollen should be classified as Symplocos (Krutzsch, 1989
). Fossil endocarps and pollen grains are known from the Paleocene to the Miocene in North America, from the Lower Eocene to the Pliocene in middle Europe, and from the Upper Eocene and the Pliocene in Japan (Kirchheimer, 1949
; Nooteboom, 1975
; Krutzsch, 1989
; Manchester, 1999
). From the fossil record, Krutzsch (1989)
has synthesized the following scenario for the historical biogeography of Symplocos. After a Late Cretaceous or Paleocene North American origin, Symplocos migrated across the North Atlantic and by the Eocene was widespread throughout the Northern Hemisphere. Range restrictions in the Miocene displaced the group southward and dispersal to South America occurred in the Pliocene. No early Tertiary fossils of Symplocos are yet known from Africa, India, Australia, or southern Southeast Asia (Krutzsch, 1989
).
Because the first-diverging lineages are all from eastern Asia, the data support an eastern Asian origin for the geographic origin of Symplocos rather than the North American origin hypothesized by Krutzsch (1989
; Fig. 4A). Our inferences are based on molecular phylogeny, whereas those of Krutzsch (1989)
are based primarily on spatiotemporal fossil data. Phylogenies based on extant taxa in which the basal nodes are exclusively eastern Asian are proving to be a predominant pattern among disjuncts between eastern Asia and eastern North America (Donoghue et al., 2001
; Xiang and Soltis, 2001
; Wen et al., 2002
). This implies a geographic origin in eastern Asia for many Northern Hemisphere disjuncts. From perspectives based on fossil data, however, it has been proposed that eastern Asia might best be considered a large refugium for many taxa of mixed-mesophytic forests once distributed widely across the Northern Hemisphere that have since become extinct in other areas (Tiffney, 1985a
, b
; Manchester, 1999
). Although the earliest Symplocos fossils are from North America, those from the Eocene of Japan suggest that the rich diversity of Symplocaceae in eastern Asia today resulted from a long and continuous presence of the family in eastern Asia throughout most of the Tertiary.
The six area optimizations of Symplocos that infer dispersal from eastern Asia to North America (followed by dispersals to the Antilles and South America) are consistent with a scenario in which one of the Northern Hemisphere high-latitude land bridges (Beringian or North Atlantic) provided a means of overland migration between eastern Asia and North America. The sole area optimization that infers dispersal from Asia to the Antilles can only be accommodated through extinction in North America, or long-distance dispersal. A pattern of dispersal from an eastern Asian origin through North America to the Antilles and South America is also the most likely migration route for the amphi-Pacific tropical group Styrax section Valvatae (Styracaceae; Fritsch, 2001
, 2003
). Like Symplocos, this is an ericalean group in which the vast majority of species are montane. The identical pattern of dispersal inferred from the phylogeny of Styrax and Symplocos suggests a common geohistorical explanation for both, and should prompt further research into the historical biogeography of other montane ericalean amphi-Pacific tropical genera (e.g., Gaultheria [Ericaceae], Saurauia [Actinidiaceae], Ternstroemia [Ternstroemiaceae]) to search for shared biogeographical patterns among these and other amphi-Pacific tropical disjuncts.
Further insight into the historical biogeography of Symplocos will require a comprehensive examination of its fossil record, more extensive taxon sampling for molecular phylogenetic data, and divergence time estimates of intercontinentally disjunct clades. Sampling should focus particularly on tropical South American species of subgenus Symplocastrum, many of which have been described since Brand's (1901)
comprehensive treatment of the genus (see, e.g., Ståhl, 1991
, 1995
). These species have not been sampled extensively, yet have the most potential to affect the outcome of biogeographic analyses of the Neotropics. Complete four-gene data are also desirable from samples of section Neosymplocos to further resolve the position of this section within the clade otherwise corresponding to section Symplocastrum.
|
|
|
FOOTNOTES |
|---|
5 Current address: Ministry of Education Key Laboratory for Biodiversity Science and Ecological Engineering, Institute of Biodiversity Science, Fudan University (IBSFU), Shanghai 200433, People's Republic of China
6 Authors for correspondence (e-mail: pfritsch{at}calacademy.org
; lssssh{at}zsu.edu.cn
)
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Anderberg A. A. C. Rydin M. Källersjö 2002 Phylogenetic relationships in the order Ericales s.l.: analyses of molecular data from five genes from the plastid and mitochondrial genomes. American Journal of Botany 89: 677-687
Angiosperm Phylogeny Group. 1998 An ordinal classification for the families of flowering plants. Annals of the Missouri Botanical Garden 85: 531-553[CrossRef][ISI]
Angiosperm Phylogeny Group. 2003 An update of the Angiosperm Phylogeny Group classification for the orders and families of flowering plants: APG II. Botanical Journal of the Linnean Society 141: 399-436[CrossRef]
Barth O. M. 1979 Pollen morphology of Brazilian Symplocos species (Symplocaceae). Grana 18: 99-107
Barth O. M. 1982 The sporoderm of Brazilian Symplocos pollen types (Symplocaceae). Grana 21: 65-69
Bidá A. 1995 Revisão taxonômica das espécies de Symplocos Jacq. (Symplocaceae) do Brasil. Ph.D. dissertation, Universidade de São Paulo, São Paulo, Brazil
Borhidi A. 1996 Phytogeography and vegetation ecology of Cuba, 2nd ed. Akadémiai Kiadó, Budapest, Hungary
Brand A. 1901 Symplocaceae. In A. Engler [ed.], Pflanzenreich IV, 242. Engelmann, Leipzig, Germany
Cronquist A. 1981 An integrated system of classification of flowering plants. Columbia University Press, New York, New York, USA
Dieffenbach C. W. G. S. Dveksler [eds.] 1995 PCR primer: a laboratory manual. Cold Spring Harbor Laboratory Press, Plainview, New York, USA
Donoghue M. J. C. D. Bell J. H. Li 2001 Phylogenetic patterns in Northern Hemisphere plant geography. International Journal of Plant Sciences 162: (6 Supplement) S41-S52[CrossRef]
Downie S. R. D. S. Katz-Downie M. F. Watson 2000 A phylogeny of the flowering plant family Apiaceae based on chloroplast DNA rpl16 and rpoC1 intron sequences: towards a suprageneric classification of subfamily Apioideae. American Journal of Botany 87: 273-292
