| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Systemics and Phytogeography |
2Missouri Botanical Garden, Graduate Studies Program, P.O. Box 299, St. Louis, Missouri 63166 USA; 3Department of Biology, Washington University, One Brookings Drive, Campus Box 1229, St. Louis, Missouri 63130-4899 USA; 4Research Center for Biodiversity, Academia Sinica, Nankang, Taipei 115, Taiwan; 5Department of Plant Biology, University of Illinois at Urbana-Champaign, Urbana, Illinois 61801 USA; 6Department of Plant Systematics and Geography, Warsaw University, Aleje Ujazdowskie 4, PL-00-478 Warszawa, Poland
Received for publication March 18, 2005. Accepted for publication August 22, 2005.
ABSTRACT
The alpine ecosystem is the only terrestrial biogeographic unit that is distributed globally. Studying phylogenetics of the plant species in this widespread ecosystem can provide insights into the historical biogeographic processes that have shaped the global biodiversity. The trans-Pacific disjunct alpine genus Oreomyrrhis (Apiaceae) was investigated using nrDNA ITS sequences to test the taxonomic and biogeographic hypotheses. Phylogenetic analyses using maximum parsimony, maximum likelihood, and Bayesian inference revealed that species of Oreomyrrhis form a weakly supported monophyletic clade that is nested within Chaerophyllum sect. Chaerophyllum (subtribe Scandicinae, tribe Scandiceae). The optimal solutions of dispersal-vicariance analysis indicate that the ancestor of Chaerophyllum sect. Chaerophyllum (including Oreomyrrhis) was distributed in Eurasia and subsequently dispersed to North America and southern Pacific Rim. Based on dating using ITS sequence variation, these dispersal events were most likely recent, probably during late Tertiary to Quaternary. The structure of the ITS haplotype network suggests that a rapid range expansion via long-distance dispersal had been crucial in generating the trans-Pacific disjunction of Oreomyrrhis. Furthermore, evolution toward smaller mericarp size and a transition from outcrossing to selfing during Oreomyrrhis's evolution might have increased the chances for long-distance dispersal, facilitating its range expansion and occupation on alpine environments.
Key Words: alpine plants • Apiaceae • Chaerophyllum • internal transcribed spacer (ITS) • long-distance dispersal • Oreomyrrhis • Pleistocene glaciations • South Pacific disjunction
The alpine ecosystem is the only terrestrial biogeographic unit that is distributed globally (Körner, 2003
). Studying phylogenetics of the plant species in this widespread ecosystem presents a unique opportunity to understand the historical biogeographic processes that have shaped global biodiversity. Recent studies in Europe suggest that dispersal dynamics, Pleistocene glaciations, and landscape heterogeneity have been the most influential factors shaping the geographic distribution and genetic variation in alpine plants (Tribsch and Stuessy, 2003
). However, few studies have attempted to understand how these historical, geological, and biological factors have affected the alpine biodiversity outside of Europe, on a global scale. The trans-Pacific disjunct genus, Oreomyrrhis, found in isolated alpine life zones across different southern latitudes around the southern Pacific Basin, presents an ideal system to test these biogeographic hypotheses.
Oreomyrrhis Endl. (Apiaceae subfamily Apioideae), as circumscribed in the seminal work of Mathias and Constance (1955)
and with novelties described subsequently by Mathias and Constance (1977)
and Chen and Wang (2001)
, consists of ca. 25 species distributed in alpine and subalpine areas and sub-Antarctic islands around the southern Pacific Rim (Mexico, Central American highlands, northern South American Andes, Tierra del Fuego, Falkland Islands, New Zealand, SE Australia, Tasmania, New Guinea, Borneo, and Taiwan; Fig. 1). It is one of the few genera in Apioideae with a mainly Southern Hemisphere distribution (Mathias, 1965
). Known as an "exemplary demonstration of the sub-Antarctic-South Pacific pattern of distribution" (van Steenis, 1963
), its unique geographic pattern has intrigued generations of phytogeographers (Hayata, 1911
; Merrill, 1918
; Mathias and Constance, 1955
; van Steenis, 1962
, 1963
, 1964
; Raven, 1973
; Melville, 1981
).
|
; Thorne, 1972
; Raven, 1973
) to ancient vicariance such as Tertiary land-bridges (van Steenis, 1962
, 1964
) and plate tectonic movements dating to the Jurassic (Melville, 1981
). These competing hypotheses predict different evolutionary relationships among Oreomyrrhis species that are readily testable with phylogenetic data. Elucidating the evolutionary history of Oreomyrrhis will also contribute to our understanding of broad patterns of Pacific and global alpine biogeography.
The taxonomic history of Oreomyrrhis began with the description of Myrrhis andicola by Kunth in 1821, based on a Humboldt and Bonpland collection from Ecuador (Humboldt et al., 1821
). Later that year, a new genus/species Caldasia chaerophylloides Lag. was described from Peru (La Gasca, 1821
). De Candolle (1829
, 1830
) accepted this new genus and transferred M. andicola to Caldasia. However, because Caldasia Lag. was a later homonym, Endlicher (1839)
proposed to replace it with the new generic name Oreomyrrhis, which has been accepted since that time (Mathias and Constance, 1955
).
Morphologically, Oreomyrrhis is characterized by an inflorescence consisting of a simple umbel borne terminally on peduncles that arise from the bases of sheathing and often rosetted leaves (Mathias and Constance, 1955
). Such a simple umbel is extremely rare in Apioideae, appearing consistently only in Oreomyrrhis, Lilaeopsis Greene, and Neogoezia Hemsl. (Constance, 1987
). The first tribal classification that included Oreomyrrhis placed the genus (as Caldasia) in tribe Scandiceae Spreng. (de Candolle, 1830
), which was challenged by Hooker (1856)
but followed by Bentham (1867)
and Koso-Poljansky (1916)
. Emphasizing the presence or absence of vittae (secretory canals) and secondary ridges on the mericarps, Bentham (1866)
suggested that, among the genera of Scandiceae, Oreomyrrhis is most similar to Chaerophyllum L. of the subtribe Scandicinae Tausch.
In one of the most comprehensive and influential taxonomic works on Apiaceae, Drude (1898)
, whose taxonomy emphasized microscopic structure of the mericarps, removed Oreomyrrhis from Scandiceae on the basis of mericarp shape and the absence of crystal druses around the carpophore (Mathias and Constance, 1955
). Drude placed Oreomyrrhis in tribe Smyrnieae Spreng., that included other simple or irregularly compound-umbelled genera, such as Neogoezia, Apiastrum Nutt. ex Torr. et A. Gray, Erigenia Nutt., and Orogenia S. Watson (Mathias and Constance, 1955
). Drude's treatment has been followed by many subsequent authors (e.g., Shan and Sheh, 1979
) and was adopted with minor changes in the most recent synopsis of Apiaceae (Pimenov and Leonov, 1993
). However, the phylogenetic position of Oreomyrrhis was left unanswered by Mathias and Constance (1955
, p. 349) as they concluded "... We are nevertheless inclined to regard Oreomyrrhis as having a vague relationship to Chaerophyllum and probably no close affinity with other genera bearing simple umbels. Although Oreomyrrhis is clearly referable to the subfamily Apioideae, we have been unable to recognize any close relative among the living Umbelliferae of either the northern or southern hemisphere."
All 13 species of Oreomyrrhis investigated to date have a base chromosome number of n = 6 (Mathias and Constance, 1955
; Moore, 1971
; Chen and Wang, 2001
; Pimenov et al., 2003
). This number is rare in Apioideae, representing only 4% of the 1461 species with published chromosome counts in the subfamily (Moore, 1971
; Pimenov et al., 2003
). The rarity of its base number has stimulated much speculation regarding Oreomyrrhis's phylogenetic position (Moore, 1971
; Raven, 1975
; Pimenov et al., 2003
). Recently, serological (Shneyer et al., 1992
) and molecular phylogenetic studies (Downie and Katz-Downie, 1996
; Downie et al., 2001
) have shown that Drude's tribe Smyrnieae is a heterogeneous and polyphyletic assemblage. Unfortunately, Oreomyrrhis has not been included in any modern systematic study and its phylogenetic affinity remains puzzling.
In addition to its dubious tribal position, the infrageneric taxonomy of Oreomyrrhis has been equally contentious (Mathias and Constance, 1955
; Stevens, 1990
). The majority of these controversies have centered on the delimitation of O. andicola (Kunth) Endl. ex Hook. f. (Mathias and Constance, 1955
). First described from the Ecuadorian Andes (as Myrrhis andicola), subsequent collections extending from the northern Andes north to the Mexican highlands indicate that O. andicola is an extremely polymorphic complex. These populations display continuous morphological variation in habit, size, leaf shape, color, and pubescence that nearly spans the range of morphological diversity of the entire genus (Mathias and Constance, 1955
; Melville, 1981
). While there have been attempts to divide the Andean Oreomyrrhis into multiple species (e.g., Urban, 1880–1882
; Johnston, 1938
), some authors have extended the species boundary of O. andicola beyond South America to include populations in New Zealand (Kirk, 1899
), Australia (Bentham, 1866
), and New Guinea (Buwalda, 1951
).
By restricting the name Oreomyrrhis andicola to encompass only the Andean populations and by accepting most regional endemics, Mathias and Constance (1955)
recognized 23 species of Oreomyrrhis (Fig. 1). Similar species recognition criteria were later adopted in Mathias and Constance (1977)
and Chen and Wang (2001)
to describe new species from New Guinea and Taiwan, respectively. However, the species delimitation by Mathias and Constance (1955)
was later criticized by Kern (1958)
as being too narrow. Kern (1958)
surmised that Oreomyrrhis comprises only a few species, with O. andicola being a species with a circum-Pacific distribution from Mexico to Taiwan. Despite Kern's criticism, Mathias and Constance's delimitations have been widely adopted in most local floras (e.g., Allan, 1961
; van Royen, 1983
; Walsh and Entwisle, 1999
). Based on the work of Mathias and Constance (1955)
, Melville (1981)
reconstructed the infrageneric relationships of Oreomyrrhis and, based on his own hypothesis of Oreomyrrhis's phylogeny, proposed a vicariance scenario that is linked with ancient geological events. Melville's phylogenetic hypothesis was later adopted by Crisci et al. (1991)
for a historical biogeographic analysis of southern South America.
In recent years, DNA sequence data have proven to be a promising and powerful tool for elucidating relationships within Apiaceae that have been difficult to resolve with morphological data alone (Downie et al., 2001
). In particular, the large quantity of data in GenBank from the nuclear ribosomal DNA (nrDNA) internal transcribed spacer (ITS) region for numerous taxa of Apiaceae are a valuable asset for testing taxonomic hypotheses surrounding Oreomyrrhis. Moreover, ITS sequence data are useful for reconstructing infrageneric relationships (e.g., Downie et al., 2004
; Neves and Watson, 2004
) and for inferring the biogeographic histories of Apiaceae genera (e.g., Wen et al., 2002
).
In an ongoing project of global alpine biogeography, Oreomyrrhis was investigated using a combination of phylogenetic, phylogeographic, and population genetic approaches. In this study, ITS sequence data were analyzed by various phylogenetic methods to address the following questions: (1) Is Oreomyrrhis monophyletic? (2) What are its closest relatives? (3) Does vicariance or long-distance dispersal better explain the present distribution of Oreomyrrhis?
MATERIALS AND METHODS
Sampling strategy for Oreomyrrhis
Species of Oreomyrrhis were sampled to optimize coverage of the geographic distribution of the genus (Fig. 1) and intraspecific morphological variation. Leaves were collected and preserved on silica gel by the first author during fieldwork in Ecuador (O. andicola), Taiwan (O. involucrata var. pubescens, O. nanhuensis, and O. taiwaniana), SE Australia (O. brevipes, O. ciliata, O. eriopoda, and O. pulvinifica), Tasmania (O. argentea, O. ciliata, O. eriopoda, and O. sessiliflora), New Zealand (O. colensoi, O. ramosa, and O. rigida), and Guatemala (O. daucifolia) and through the courtesy of local researchers in Taiwan (O. involucrata and O. taiwaniana), Tierra del Fuego (O. hookeri), the Falkland Islands (O. hookeri), and Mexico (O. orizabae). DNA was also extracted from herbarium specimens for some taxa (O. azorellacea, O. borneensis, O. gunnii, O. involucrata, O. linearis, O. papuana, O. pumila, and O. sessiliflora), including two of the most unique New Guinea species, O. azorellacea (moss-like cushion plant) and O. linearis (characterized by grass-like linear leaves). The species not sampled in this study include the rare New Guinean species O. buwaldiana and O. plicata, that are known only by type collections, and the Mexican high alpine species O. tolucana for which no ITS sequence was amplified after several trials using herbarium material.
To evaluate the monophyly of species with broad distributions and substantial morphological variation, multiple individuals (number in parentheses) from different localities were sampled for O. andicola (5), O. argentea (2), O. ciliata (5), O. colensoi (4), O. eriopoda (6), O. hookeri (2), O. involucrata (4), O. linearis (2), O. ramosa (4), O. rigida (4), O. sessiliflora (2), and O. taiwaniana (4). Oreomyrrhis involucrata var. pubescens, a taxon not recognized by Mathias and Constance (1955)
and Chen and Wang (2001)
but readily recognizable in the field (K. Chung, personal observation), was also surveyed. One individual from Awahokomo karstland, Otago, New Zealand, previously identified as O. rigida (Molloy et al., 1999
), but likely a new species (P. Heenan, Landcare Research, personal communication), was also investigated. In total, 57 accessions of Oreomyrrhis, representing 22 of the 25 species recognized and/or later described by Mathias and Constance (1955
, 1977
) and Chen and Wang (2001)
, were included. The taxa sampled, taxon authorities, voucher information, and Genbank accession numbers are listed in the Appendix. Throughout this study, Mathias and Constance's (1955)
species circumscription was closely followed for plant identification, with further verification based on examination of type specimens and/or specimens annotated by Mathias and Constance (1955)
.
DNA extraction, PCR amplification, and sequencing
Total genomic DNA was isolated from silica-dried leaf material using the CTAB procedure of Doyle and Doyle (1987)
. For herbarium specimens, DNA was isolated using a VIOGENE Plant Genomic DNA extraction miniprep system (Viogen U.S.A., Sunnyvale, California) following the manufacturer's protocol. The entire ITS region (ITS-1, 5.8S, and ITS-2) was amplified using primers ITS1 and ITS4 described in White et al. (1990)
. For older herbarium material, the internal primers ITS2 and ITS3 (Downie and Katz-Downie, 1996
) or primers ITS-C and ITS-D (Blattner, 1999
), in combination with the external primers ITS1 and ITS4, were used to amplify the entire ITS region. All amplifications were performed in a 50-µL reaction. Polymerase chain reaction (PCR) conditions were set for an initial 4 min denaturation at 94°C, followed by 35 cycles of 94°C for 30 s, 50°C for 30 s, and 72°C for 1 min, with a final extension at 72°C for 10 min and a final incubation at 30°C for 10 min. PCR products (templates) were visualized via agarose gel electrophoresis and were further purified using the QIAquick Gel Extraction Kit (Qiagen, Valencia, California). Sequence reactions were carried out in both directions for each purified PCR template using the ABI Prism dGTP BigDye Terminator Ready Reaction kit (Applied Biosystems, Foster City, California, USA) following the manufacturer's protocol. Excessive Bigdye in the sequence reactions was removed using CENTRI-SEP columns (Princeton Separations, Adelphia, New Jersey, USA). Sequence reaction products were visualized using either an Applied Biosystem 373 automated DNA sequencer or a BaseStation DNA Fragment analyzer (MJ Research, South San Francisco, California, USA). Contigs were assembled using SEQMAN (Swindell and Plasterer, 1997
). All newly acquired sequences have been archived in Genbank (Appendix).
Molecular cloning
To examine the extent of sequence homogenization among reiterated ITS copies, molecular cloning was conducted on 10 accessions of nine Oreomyrrhis species (O. andicola [Chung 1493], O. daucifolia [Chung 1650], O. eriopoda [Chung 1554 and Chung 1612], O. hookeri [D. Broughton, no voucher], O. involucrata [Juan 53], O. nanhuensis [Chung 1510], O. orizabae [Avendaño R. 5349], O. ramosa [Chung 1630], and O. rigida [Chung 1618]). The purified ITS templates were ligated to the pGEM-T Easy vector (Promega, Madison, Wisconsin, USA) following the manufacturer's protocol and subsequently transformed into competent cells. After an overnight culture at 37°C on the LB ampicillin/IPTG/X-gal selective plate, colonies carrying the ITS insert were identified by color (white) and further verified by PCR using the T7 and SP6 promoter primer pairs (Promega, Madison, Wisconsin, USA). For each accession, up to 20 white colonies were checked for inserts. The amplified PCR products were purified and subsequently sequenced using the ITS1 and ITS4 primers as described in the previous section.
Tests of sequence evolution
The rate constancy of the ITS sequence evolution of Oreomyrrhis and its closest relatives Chaerophyllum procumbens and C. tainturieri (see Figs. 3 and 4) were inferred from relative rate tests (Wu and Li, 1985
) based on Kimura two-parameter distance (K2P; Kimura, 1980
), implemented by the program K2WuLi (Jermiin, 1996
). Chaerophyllum temulum was used as a reference taxon. Given a constant ITS evolution rate, approximate times of diversification with 95% confidence intervals between Oreomyrrhis and its outgroup and among Oreomyrrhis species were estimated as ± 2 SE of the K2P distances, calibrated with an ITS sequence divergence rate of 0.79% per million years (Sang et al., 1994
; Kropf et al., 2002
). This more conservative ITS divergence rate was chosen because we were mainly interested in testing competing biogeographic scenarios related to recent (late Tertiary-Quaternary) vs. old (Jurassic or early Tertiary) diversification of Oreomyrrhis, as predicted by long-distance dispersal (e.g., Dawson, 1971
; Raven, 1973
) vs. vicariance hypotheses (Melville, 1981
; van Steenis, 1964
), respectively. Pairwise K2P distances and their SEs were calculated using MEGA2 (Kumar et al., 2001
).
|
|
Analysis 1
To elucidate Oreomyrrhis's phylogenetic position within Apioideae, the data matrix of Downie and Katz-Downie (1996)
was adopted with some modifications. This matrix was composed of ITS-1 and ITS-2 sequences of representatives of tribe Smyrnieae, subtribes Scandicinae, Daucinae, and Torilidinae of tribe Scandiceae, and several major clades of Apioideae identified by recent molecular systematic studies (Downie et al., 2001
). Pleurospermum foetens Franch. (tribe Symrnieae, sensu Drude [1898]
) was chosen as outgroup (cf. Downie et al., 2001
). ITS sequences of eight Oreomyrrhis species from different geographic regions [O. orizabae (Mexico), O. andicola (South American Andes), O. hookeri (Tierra del Fuego), O. colensoi (New Zealand), O. eriopoda (Australia), O. papuana (New Guinea), O. borneensis (Borneo), and O. taiwaniana (Taiwan)] were added to the matrix. Chaerophyllum temulum L., the type species of the largest genus of Scandiceae (Spalik and Downie, 2001
), was added to test Bentham's (1866)
proposition that Oreomyrrhis is most similar to Chaerophyllum. Other additions included selected taxa of Drude's (1898)
tribe Smyrnieae that possess irregularly compound umbels (Erigenia, Neogoezia, and Orogenia), and Lilaeopsis, another predominantly Southern Hemispheric distributed genus also possessing a simple umbel (Constance, 1987
; Petersen et al., 2001
). Sequences were aligned using DAMBE (Xia and Xie, 2001
) and were manually adjusted in the alignment editor Se-Al (Rambaut, 1996
). Only those positions that were aligned unambiguously were used in the phylogenetic analyses; the aligned data matrix is available in TreeBASE (study accession number S1350, matrix accession number M2384).
Phylogenetic analyses were conducted using maximum parsimony (MP) algorithm implemented in PAUP* version 4.0b10 (Swofford, 2003
). All characters were treated as unordered, and all character transformations were weighted equally. A heuristic search with 1000 random addition replicates was performed with the following options in effect: MULTREES, TBR branch swapping, and gaps treated as missing data. Clade support was assessed by bootstrap (Felsenstein, 1985
) and decay (Bremer, 1988
) analyses. One hundred bootstrap replicates were analyzed by PAUP* using the heuristic search option with 10 random sequence additions. Decay indices were calculated using the program TreeRot.v2 (Sorenson, 1999
). To evaluate the phylogenetic content of the data, the g1 statistic of tree-length distribution skewness was calculated from 10 000 random MP trees generated by PAUP*, and was subsequently compared to the critical values provided in Hillis and Huelsenbeck (1992)
.
Analysis 2
The results of analysis 1 unambiguously placed Oreomyrrhis in subtribe Scandicinae of tribe Scandiceae, with Chaerophyllum being its closest relative (Fig. 2). To further elucidate their relationships within Scandicinae, ITS sequences of 57 Oreomyrrhis accessions (22 species) were combined with all available ITS sequences (ITS-1 and ITS-2) of the genus Chaerophyllum (22 species) and representatives of other major clades in the subtribe (Anthriscus, Kozlovia, Myrrhis, Osmorhiza, Scandix, Sphallerocarpus; see Spalik and Downie, 2001
). Athamanta cretensis and Conopodium ramosum were selected as outgroups based on recent phylogenetic analysis (Downie et al., 2000
). Voucher information of sampled taxa is shown in the Appendix.
|
. Both matrices were analyzed using the same criteria described in analysis 1. Bootstrap support values, decay indices, and the g1 statistic for the gap-excluding matrix were calculated based on methods in analysis 1.
The gap-excluding data matrix was then analyzed using maximum likelihood (ML). Modeltest 3.5 (Posada and Crandall, 1998
) was used to identify the appropriate evolutionary model of nucleotide substitution. Based on the model selected by Modeltest 3.5, 10 heuristic searches were performed in PAUP* using random addition sequence and TBR branch swapping. To assess clade support, a bootstrap analysis of 1000 replicates was conducted using the neighbor-joining method with ML distance (e.g., Downie et al., 2004
; Sun et al., 2004
).
A Bayesian phylogenetic analysis was also performed using the program MrBayes 3 (Ronquist and Huelsenbeck, 2003
) based on the same nucleotide substitution model used in ML analysis. In MrBayes, parameters were set as follows: random starting tree, 2 000 000 generations runs with sampling occurring every 100 generations, and a 4000-tree "burn-in" period. PAUP* was used to generate a 50% majority consensus tree of the output tree file from MrBayes.
Analysis 3
Results of analysis 2 revealed that Oreomyrrhis accessions formed a poorly resolved polytomous clade probably resulting from low sequence divergences. Such levels of sequence variation suggest ongoing speciation processes in Oreomyrrhis. To better elucidate relationships among Oreomyrrhis accessions, a haplotype network was constructed. Theoretically, haplotype networks can straddle population–species interfaces where speciation occurs, providing a powerful tool to investigate evolutionary processes (Templeton et al., 2000
; Schaal and Leverich, 2001
). We compiled an ITS sequence (ITS-1, 5.8S, and ITS-2) matrix of 57 Oreomyrrhis accessions. Based on the results of analysis 2 (see Figs. 3 and 4), Chaerophyllum procumbens and C. tainturieri were selected as outgroups. The haplotype network was calculated using statistical parsimony (Templeton et al., 1992
) in the program TCS (Clement et al., 2000
), with alignment gaps treated as a fifth state.
Biogeographic analysis
A dispersal-vicariance analysis (DIVA; Ronquist, 1997
) was performed to infer biogeographic history of Oreomyrrhis and subtribe Scandicinae. Given a phylogeny, DIVA reconstructs distribution of ancestral areas that minimize dispersal and extinction events under a parsimony criterion (Ronquist, 1997
), presenting an ideal tool to identify sources and directions of dispersal for Oreomyrrhis and its close relatives. For implementing DIVA, a fully resolved phylogeny is demanded. Because parts of the Scandicinae phylogeny are unresolved, clades composed of taxa distributed in the same general unit area were compressed to generate a simplified bifurcating phylogeny that is required by DIVA. Information of taxon distribution of subtribe Scandicinae was gathered from Spalik and Downie (2001
; Fig. 3). Athamanta and Conopodium were chosen as outgroups (Spalik et al., 2001a
). Since we were mainly interested in identifying the ancestral source of dispersal for Oreomyrrhis, not the exact locality of initial diversification, the unresolved clade of Oreomyrrhis (Fig. 4A) was also compressed to a single terminal and designated with a Pacific distribution (D). Three additional general unit areas were defined to cover the distributional range of Scandicinae: Eurasia (including North Africa; A), North America (B), and South America (C).
RESULTS
Characteristics and evolution of Oreomyrrhis ITS sequences
From 57 accessions of 22 Oreomyrrhis species, 26 distinct ITS sequences (haplotypes) were obtained. The alignment of the polymorphic sites, haplotype designations, and their frequencies are shown in Table 1. The length of the entire ITS region for the 26 Oreomyrrhis haplotypes ranged from 602 to 607 bp (ITS-1: 217–218 bp; 5.8S: 163–164 bp; ITS-2: 221–226 bp). Alignment of the 26 haplotypes resulted in a matrix of 610 positions (ITS-1: 219 bp; 5.8S: 164 bp; ITS-2: 227 bp), including two 1-bp gaps in ITS-1, a single 1-bp gap in 5.8S, and three 1-bp gaps and a 4-bp gap in ITS-2.
|
) were conducted to evaluate if intra-individual paralogous ITS sequences are present and shared with other species and thereby would mislead phylogenetic inferences. These ITS clones were added to the matrices used in analyses 1, 2, and 3, respectively. Phylogenetic as well as haplotype network reconstructions resulted in congruent tree topologies as the ITS clones were excluded (data not shown). In particular, ITS clones within each accession coalesced to the dsITS, suggesting that paralogous ITS copies within each accession originated after speciation, a condition of sequence homology known as "shallow paralogy" (Bailey et al., 2003
). Because shallow paralogy does not interfere with the phylogeny reconstruction (Hershkovitz et al., 1999
; Bailey et al., 2003
), only dsITS were used in all subsequent phylogenetic analyses.
Within the 378 relative rate tests among C. procumbens, C. tainturieri, and the 26 Oreomyrrhis ITS haplotypes, there were only two tests where rate constancy of ITS sequence evolution was rejected significantly (P < 0.05; z score > |1.96|). This figure is much lower than the expected number by chance alone (one of every 20 comparisons will be rejected significantly at the 5% level by chance; Sytsma and Schaal, 1985
; Abbott and Comes, 2004
). Given the apparent rate constancy of ITS evolution, times of speciation and diversification were approximated based on pairwise sequence divergences, calibrated with an ITS sequence divergence rate of 0.79% per million years. The average pairwise sequence divergence between C. temulum and Oreomyrrhis and its closest relatives (C. procumbens and C. tainturieri) was 6.45 ± 1.18%, corresponding to a separation time of 8.16 (95% confidence intervals = 5.18–11.15) million years ago (mya). Between Oreomyrrhis and its closest relatives (C. procumbens and C. tainturieri), the average pairwise sequence divergence is 1.76 ± 0.49%, indicating that Oreomyrrhis diverged from its most recent common ancestor at 2.23 (95% confidence intervals = 0.99–3.47) mya. Within Oreomyrrhis, the average pairwise sequence divergence is 0.80 ± 0.18%, corresponding to a diversification time of 1.01 (95% confidence intervals = 0.56– 1.47) mya.
Analysis 1
Alignment of the ITS-1 and ITS-2 sequences of the seven Oreomyrrhis ITS sequences (O. hookeri and O. orizabae are identical) of eight Oreomyrrhis species and 44 representatives of Apioideae resulted in a data matrix of 475 positions. Due to ambiguous alignment, 42 characters (70–74, 102–104, 249–258, 268–285, 388–389, and 409–412) were excluded. The final 433-bp matrix contained 264 parsimony-informative characters. Parsimony analysis resulted in 144 equally most parsimonious trees (MPTs) of 1218 steps. The uncorrected consistency index (CI) was 0.48 and the corrected CI was 0.44, the latter of which was higher than the expected CI (0.31) for 52 taxa (Sanderson and Donoghue, 1989
). The retention index (RI) was 0.72. The strict consensus tree of the 144 MPTs with bootstrap support values (BV) and decay indices (DI) for each clade is presented in Fig. 2. The g1 statistic for 10 000 random trees was –0.39, which is significantly more skewed than random data (g1 = –0.09 for 250 variable position and 25 or more taxa; P < 0.01; Hillis and Huelsenbeck, 1992
).
Compared to the results of Downie and Katz-Downie (1996)
, whose sampling strategy was basically adopted here, relationships among taxa that were both sampled by previous and current studies are highly concordant (Fig. 2). One of the major differences was the placement of Smyrnium olusatrum, which was placed sister to tribe Scandiceae with strong support in Downie and Katz-Downie (1996)
but was placed next to tribe Oenantheae with weak support in the current study (Fig. 2).
The eight Oreomyrrhis species formed a well-supported monophyletic group (BV = 93%; DI = 3) with low internal resolution (Fig. 2). This Oreomyrrhis clade was sister to Chaerophyllum temulum, and together they formed a strongly supported clade (BV = 99%; DI = 8) within subtribe Scandicinae. None of the genera that were assigned to tribe Smyrnieae by Drude (1898)
and other genera with simple and/or irregular compound umbels (Fig. 2) are closely related to Oreomyrrhis.
Analysis 2
Alignment of 26 Oreomyrrhis haplotypes, 22 Chaerophyllum species (sequences of Chaerophyllum elegans/ C. hirsutum and C. hakkiaricum/C. macrospermum are identical, respectively), and 11 outgroups resulted in a matrix of 57 terminals and 452 characters. Twenty-nine gaps of various sizes (nineteen 1-bp, four 2-bp, one 3-bp, two 4-bp, one 8-bp, one 11-bp, and one 16-bp) were introduced to facilitate alignment. Ten of these gaps were parsimony informative (gap A: position 68; B: 72; C: 101; D: 237, E: 255; F: 375–377; G: 411; H: 247–262; I: 266–269; J: 344–345; Fig. 3).
Treating gaps as missing data, MP analysis of the ITS data matrix resulted in 2680 MPTs of 463 steps (CI = 0.70, corrected CI = 0.62; RI = 0.84). The g1 statistic for 10 000 random trees was –0.46, which is significantly more skewed than that of random data (Hillis and Huelsenbeck, 1992
). To test the phylogenetic utility of the alignment gaps, parsimony-informative gaps were scored as binary characters, resulting in a second matrix of 462 aligned positions. Maximum parsimony analysis revealed 1580 MPTs of 490 steps with slightly lower consistency and retention indices (CI = 0.68, corrected CI = 0.60; RI = 0.83), suggesting an increase in homoplasy when gaps were taken into account. The strict consensus trees derived from the two matrices (with and without gap characters), however, were almost identical, except for two additional clades (denoted by asterisks in Fig. 3) uncovered in the gap-including matrix. The strict consensus tree derived from the gap-including matrix is presented in Fig. 3. Mapping the 10 parsimony informative gaps onto the consensus tree indicated that only three gaps were free of homoplasy (Fig. 3).
Modeltest 3.5 selected the GTR + G model (base frequencies: 0.2403, A; 0.2182, C; 0.2480, G; 0.2935, T; estimates of substitution rates A
C: 1.3354; AG: 2.3532; AT: 1.4904; CG: 0.4394; CT: 3.1987; GT: 1; proportion of invariable sites = 0; gamma distribution shape parameter = 0.9001). Based on these parameters, a single maximum likelihood (ML) tree was uncovered (–Ln likelihood score = 3153.10211; Fig. 4A). The same model and parameters were used to perform a Bayesian analysis. The 50% majority rule consensus tree with Bayesian posterior probabilities (PP) is shown in Fig. 4B.
Topologies uncovered from the three phylogenetic inference methods were very similar, with only minor differences in resolution between some terminal taxa (Figs. 3 and 4). In all analyses, monophyly of the Chaerophyllum-Oreomyrrhis clade was well to strongly supported (BV = 85% in MP and 91% in ML; DI = 3; PP = 99%). Within this clade, four strongly supported clades were uncovered. Three of these corresponded perfectly to Chaerophyllum sect. Chrysocarpum Spalik & S. R. Downie, sect. Dasypetalon Neilr., and sect. Physocaulis DC. [a monotypic section composed of only C. nodosum (L.) Crantz], respectively, that were identified by recent molecular systematic studies (Spalik and Downie, 2001
; Spalik et al., 2001a
). Section Chaerophyllum, however, was rendered paraphyletic by Oreomyrrhis.
In all analyses, Oreomyrrhis was placed within Chaerophyllum sect. Chaerophyllum with moderate to strong support (BV = 76% in MP and 88% in ML; DI = 2; PP = 99%). Within this clade, all 26 Oreomyrrhis accessions and the two North American Chaerophyllum species (C. procumbens and C. tainturieri) formed a strongly supported subclade (BV = 88% in MP and 93% in ML; DI = 2; PP = 90%) that was sister to C. temulum. Relationships within the subclade, however, were polytomous, with four moderate to strongly supported clades nested within it. These four clades are the North American Chaerophyllum clade (BV = 99% in MP and 98% in ML; DI = 5; PP = 100%), the New Zealand Oreomyrrhis clade (BV = 80% in MP and 88% in ML; DI = 2; PP = 99%), the East Asian Oreomyrrhis clade (BV = 74% in MP and 76% in ML; DI = 1; PP = 96%), and the Australian endemic clade consisting of several Australian Oreomyrrhis taxa (BV = 60% in MP and 59% in ML; DI = 1; PP = 89%).
In the ML analysis (Fig. 4A), all Oreomyrrhis accessions were grouped together in a weakly supported monophyletic group (BV = 57%) that was sister to the North American Chaerophyllum clade. However, supports for the monophyletic Oreomyrrhis were weak in both MP and Bayesian analysis (BV = 36% in MP; PP = 38%). Whether or not the Oreomyrrhis species that were represented by multiple accessions were monophyletic was inconclusive.
Analysis 3
Alignment of the 57 Oreomyrrhis accessions and the two North American Chaerophyllum species resulted in a matrix of 611 positions (Table 1). Alignment position 568 was excluded due to its potential homoplasy (Table 1; see Discussion). With alignment gaps treated as a fifth state, a single, fully resolved network was reconstructed (Fig. 5). The ITS network has a starlike topology, within which the most frequent haplotype (common Oreomyrrhis haplotype) is linked to the outgroup taxa (Chaerophyllum clade) and connected to 12 relatively low frequency haplotypes and clades, including the New Zealand clade, the East Asian clade, and the Australian clade that were also recovered in analysis 2 (Figs. 3 and 4). For species represented by multiple haplotypes (O. argentea, O. ciliata, O. colensoi, O. eriopoda, and O. ramosa), these constituent haplotypes were not grouped into separate haplotype lineages (Fig. 5).
|
; Spalik et al., 2001a
; Wen et al., 2002
) and results of the current study, a simplified phylogeny composed of 14 terminals (Fig. 6) was generated. The optimal solution of DIVA indicates that the ancestor of Scandicinae was distributed in Eurasia and four dispersal events were required to explain the current biogeographic pattern of this subtribe (Fig. 6). For our focal taxa (Chaerophyllum-Oreomyrrhis), three equally parsimonious biogeographic scenarios were obtained (Fig. 6). In all three scenarios, Eurasia was shown to be the ancestral area, from which two dispersal events to North America and Pacific were required (Fig. 6). Two scenarios suggest that the dispersal events were sequential, with either North America or the Pacific region each being equally probable of being reached first. Alternatively, the third scenario suggests two independent dispersal events to North America and Pacific, with their order unknown.
|
Utilization of nrDNA ITS region for phylogenetic inference
The properties of ITS loci that have attracted and facilitated their application in plant phylogenetic studies (Baldwin et al., 1995
) were recently reexamined by Álvarez and Wendel (2003)
. Although ITS loci are generally assumed to be homogenized by concerted evolution, Álvarez and Wendel (2003)
cautioned that, under circumstances such as hybridization, polyploidization, or pseudogene formation, separate ITS sequences might be present and persistent in a genome. This could interfere with sequencing, distort assessment of sequence homology, and subsequently mislead phylogenetic inference.
The ITS region was chosen primarily because it is the only molecular marker for which a comprehensive sampling of Apiaceae (Downie et al., 2000
, 2001
) is available for testing taxonomic hypotheses regarding the tribal placement of Oreomyrrhis. The ITS region is also the fastest evolving sequence known in Apiaceae (Downie et al., 2001
) and therefore is more likely to resolve infrageneric relationships. In Oreomyrrhis, a low base chromosome number precludes its origin through polyploidization. The autogamous nature (discussed later) and currently isolated distribution of most Oreomyrrhis species assure that interspecific hybridization and introgression are probably rare. Plant species characterized by these biological attributes are frequently assumed to have had their ITS paralogues homogenized by concerted evolution (Denduangboripant and Cronk, 2000
).
However, molecular cloning of all 10 selected Oreomyrrhis accessions revealed intra-individual ITS polymorphisms that were imperceptible from the electropherogram of direct sequencing. The persistence of intra-individual ITS polymorphism in Oreomyrrhis suggests that concerted evolution might not have proceeded fast enough to homogenize different ITS sequence types (Denduangboripant and Cronk, 2000
; Hughes et al., 2002
). One possible scenario for the failure of concerted evolution in this non-hybrid diploid genus is the existence of multiple nrDNA arrays in the genome (O'Kane et al., 1996
), which can be detected by fluorescent in situ hybridization (FISH) and genomic in situ hybridization (GISH) (Álvarez and Wendel, 2003
). Alternatively, the existence of different ITS paralogues within each accession of Oreomyrrhis may suggest that intra-individual ITS polymorphism might be more widespread than previously assumed (Hughes et al., 2002
). Nevertheless, since the revealed intra-individual ITS polymorphisms present shallow paralogy that does not impede phylogenetic inference, ITS sequences remain a credible molecular marker for phylogenetic reconstruction in Oreomyrrhis.
Phylogenetic affinities of Oreomyrrhis
The ITS data presented here unambiguously placed Oreomyrrhis in the subtribe Scandicinae of tribe Scandiceae (Fig. 2), supporting the taxonomic hypotheses of de Candolle (1830)
, Bentham (1867)
, and Koso-Poljansky (1916)
, and rejecting Drude's (1898)
placement of Oreomyrrhis in tribe Smyrnieae. Drude's tribe Smyrnieae (encompassing Arracacia, Conium, Erigenia, Neogoezia, Orogenia, Oreomyrrhis, Pleurospermum, and Smyrnium) is polyphyletic in the present analysis (Fig. 2), confirming relationships suggested by previous studies using serological (Shneyer et al., 1992
) and molecular data (Downie and Katz-Downie, 1996
; Katz-Downie et al., 1999
). Our ITS phylogenies also indicate that taxa characterized by simple or irregular compound umbels (e.g., Oreomyrrhis, Orogenia, Neogoezia, Lilaeopsis, and Erigenia) do not form a monophyletic group (Fig. 2), indicating that this trait has had multiple independent origins in Apioideae and that its phylogenetic utility has been overemphasized.
The close relationship between Oreomyrrhis and Chaerophyllum proposed by Bentham (1866)
was verified by our ITS data (Figs. 2– 4). Chaerophyllum is the largest (ca. 30 species) and most diverse genus in subtribe Scandicinae. It is distributed mainly in Eurasia (including North Africa) and is most diversified in the Mediterranean and the Caucasus regions, with two species native to North America (Spalik and Downie, 2001
). Recent molecular studies support the monophyly of Chaerophyllum and reveal four well-supported clades within it (Spalik and Downie, 2001
). Lacking morphological synapomorphies, these four clades were each treated as section by Spalik and Downie (2001)
.
Based on the ITS data, Oreomyrrhis is nested within the sect. Chaerophyllum (Figs. 3 and 4) that comprises one species native to Europe (C. temulum) and two to North America (C. procumbens and C. tainturieri). Within this section, Oreomyrrhis is most closely related to the two North American species. The phylogenetic relationships revealed by the current data indicate the paraphyletic nature of Chaerophyllum if Oreomyrrhis is treated as a separate genus. Because C. temulum is the type species of the genus, preservation of the generic status of Oreomyrrhis would simultaneously mandate the erection of four small genera (the North American Chaerophyllum clade, sect. Chrysocarpum, sect. Dasypetalon, and the monotypic sect. Physocaulis), leaving Chaerophyllum a monotypic genus composed of only C. temulum. Before proposing formal nomenclatural changes, further study by employing additional molecular markers of chloroplast genome will be conducted to test the result reached by ITS data.
Monophyly of Oreomyrrhis
The monophyly of Oreomyrrhis was only weakly supported in the ML analysis (BV = 57%; Fig. 4A). The bootstrap support in MP analysis and Bayesian posterior probability for this clade are both less than 50%. Examination of the aligned ITS matrix revealed a likely homoplasious character at position 409 (position 568 in Table 1 and analysis 3). To evaluate how this character had affected phylogeny reconstruction, we excluded this character and performed a MP analysis. Using the same parameters for the gap-excluding matrix in analysis 2, a heuristic search resulted in 536 equal length MPTs of 458 steps, two steps shorter than the original trees (460 steps). We used the "fast stepwise addition" option of PAUP* to perform a bootstrap analysis with 100 000 replicates (e.g., Sun et al., 2004
). The strict consensus of the 536 MPTs was almost identical to the ML tree (Fig. 4A), and the support for the monophyly of Oreomyrrhis accessions increased to 66%, compared to 36% when character 568 was included.
Although the ITS data provide only weak supports, the monophyly of Oreomyrrhis is unambiguously defined by its morphological synapomorphy (simple umbel) and apparent cytological and ecological distinctiveness. The monophyly indicates a single origin of Oreomyrrhis, suggesting that the trans-Pacific disjunction of the genus reflects a genuine biogeographic process, not a taxonomic artifact.
Biogeographic implications of the ITS phylogenies
Because a great proportion of the distribution range of Oreomyrrhis parallels the Gondwanan disjunction in the South Pacific (e.g., Sanmartín and Ronquist, 2004
), this genus had often been assumed to have a Southern Hemispheric and/or ancient origin (e.g., van Steenis, 1963
; Dawson, 1963
; Raven, 1973
; Melville, 1981
; Smith, 1986
; Wu, 1998
). However, our ITS data strongly support that Oreomyrrhis is closely related to Chaerophyllum of subtribe Scandicinae, that is mainly distributed in the Northern Hemisphere. Within Chaerophyllum, Oreomyrrhis is nested within sect. Chaerophyllum and, together with the two North American species of the section (C. procumbens and C. tainturieri), forms a clade that is sister to the European species C. temulum (Figs. 3 and 4). Spalik and Downie (2001)
hypothesized that Chaerophyllum arrived in North America by the incidental dispersion of seeds from Europe, probably by vagrant birds. Based on the optimization of DIVA, the ancestral area for sect. Chaerophyllum (including Oreomyrrhis) was suggested to be Eurasia, and the disjunction in North American and Pacific involved two dispersal events (Fig. 6). Because the distribution range of Oreomyrrhis and Chaerophyllum do not overlap, DIVA optimization indicates that the current distribution could have resulted from dispersal from Europe to North America followed by a second dispersal to Pacific, to Pacific first and second dispersal to North America, or two independent dispersals to Europe and Pacific (Fig. 6).
Based on our time estimates, these dispersal events probably took place around late Miocene to early Pliocene ([5.18–] 8.16 [–11.15] mya), when dispersals from Europe to North America through island hopping via the North Atlantic Land Bridge (Tiffney and Manchester, 2001
) were still possible. However, since the basal lineage of Oreomyrrhis cannot be identified confidently by our data, the initial dispersal event that gave rise to subsequent Oreomyrrhis diversification remains unclear.
The estimated times for the separation of the North American Chaerophyllum and Oreomyrrhis (2.23 mya [95% confidence intervals = 0.99–3.47 mya]) and diversification within Oreomyrrhis (1.06 mya [95% confidence intervals = 0.56– 1.47 mya]) are relatively recent, corresponding to a late Tertiary to Quaternary evolution of the group. Note that the 0.79% sequence divergence rate is at the lower end of the range reported for flowering plants (Zhang et al., 2001
), indicating that the dates inferred here might have been overestimated and the actual times for Oreomyrrhis diversification might be more recent. Even under this more conservative estimate, our results strongly support a late Tertiary to Quaternary origin and diversification of Oreomyrrhis, and most likely long-distance dispersal as the major biogeographic mechanism underlying the trans-Pacific disjunction of Oreomyrrhis (Dawson, 1971
; Thorne, 1972
; Raven, 1973
). The vicariant scenarios that invoke early Tertiary land-bridge (van Steenis, 1962
, 1964
) or plate tectonic movements dating back to the Jurassic (Melville, 1981
) can be rejected confidently because their predicted temporal contexts significantly predate the origin of Oreomyrrhis inferred from the ITS data.
Recent phylogenetic studies suggest that long-distance dispersal underlies the distributions of many trans-Pacific disjunct plant groups (Winkworth et al., 2002
; Sanmartín and Ronquist, 2004
). Comparisons of these Pacific taxa have revealed concordant phylogenetic relationships, suggesting that recent long-distance dispersal events might not occur randomly but have followed certain dispersal routes, probably associated with ancient vegetation types that may have served as migratory corridors (Winkworth et al., 2002
; Sanmartín and Ronquist, 2004
). Unfortunately, the ITS data contain only limited information regarding the pattern of dispersal that have shaped the trans-Pacific disjunction of Oreomyrrhis.
Biogeographic implications of the ITS haplotype network
To better understand genealogical processes that underlying the Pacific disjunction, an ITS haplotype network was constructed (Fig. 5). The predominant feature of the network is a single widespread and high frequency haplotype (common Oreomyrrhis haplotype) linked to numerous low-frequency haplotypes by a few mutational steps, a pattern that is commonly seen in temperate taxa in Europe and North America that have experienced drastic distributional changes during the Pleistocene (Hewitt, 1996
; Schaal et al., 1998
; Milá et al., 2000
). This type of network structure could have resulted from (1) a postglacial range expansion from refugium populations that had experienced genetic bottleneck, (2) gene flow, or (3) a selective sweep (Milá et al., 2000
). In Oreomyrrhis, trans-Pacific gene flow seems an unlikely explanation for the wide distribution of the common Oreomyrrhis haplotype (Abbott and Comes, 2004
). Because the ITS region is known to subject little functional constraint (Baldwin et al., 1995
), unless this region is closely linked to a gene favored by natural selection, a selective sweep is a less likely explanation than range expansion (Milá et al., 2000
).
However, unlike low-elevation temperate taxa whose distribution ranges expanded after the retreat of the ice sheet (Hewitt, 1996
), alpine plants such as Oreomyrrhis would have actually experienced postglacial range contraction, as rising global temperatures forced alpine plants to migrate up toward mountain tops (Raven, 1973
; Larena et al., 2002
). During the glacial maxima, alpine plant populations would have actually expanded as the alpine zone moved downward. This reverse scenario of glacial population expansion and interglacial population contraction is known as the "displacement refugia model" (Kropf et al., 2003
). Under this model, the common Oreomyrrhis haplotype likely achieved its broad distribution by "glacial" range expansion as more suitable alpine habitats became available during glacial maxima. Since the common Oreomyrrhis haplotype is the only haplotype distributed in more than one geographic region, its glacial range expansion probably had occurred before other haplotypes arose.
Species circumscription and speciation in Oreomyrrhis
Although many of the species circumscribed by Mathias and Constance (1955)
and Chen and Wang (2001)
are characterized by distinct ITS haplotypes (e.g., O. azorellacea, O. borneensis, O. brevipes, O. gunnii, O. linearis, O. nanhuensis, and O. papuana), suggesting that they have evolved as separate evolutionary lineages (Templeton et al., 2000
), such genetic distinctness is absent in the majority of the species investigated (Fig. 5). For instance, the common Oreomyrrhis haplotype is carried by nine species that are distributed throughout most of the range of Oreomyrrhis (Fig. 5). The sharing of haplotypes by recently differentiated species could result from interspecific hybridization or incomplete lineage sorting of ancestral haplotypes that had been present prior to the divergence of the species (Olsen and Schaal, 1999
). In the current study, the ancestral position of the common Oreomyrrhis haplotype suggested by its high mutational connectedness and high frequency (Castelloe and Templeton, 1994
) strongly favors lineage sorting, for interspecific hybridizations between disjunct taxa across Pacific seem unlikely (Abbott and Comes, 2004
). On the other hand, gene flows among species in the same geographic area (e.g., O. argentea A, O. eriopoda B, and O. pulvinifica in SE Australia; O. involucrata and O. taiwaniana in Taiwan; O. colensoi A, O. ramosa A, and O. rigida in New Zealand) are possible for haplotype sharing, especially during glacial maxima when alpine plant populations underwent range expansion into lower elevations (Larena et al., 2002
). Furthermore, the recency of species divergences within Oreomyrrhis suggests that there probably has not been enough time for achieving reciprocal monophyly of the newly evolved lineages (Avise, 2000
). The preliminary results from ITS data provide an exciting first step toward an in-depth understanding of the speciation in Oreomyrrhis. In the ongoing phylogeographic and population genetic studies, we are employing chloroplast sequence data and AFLP markers to explore the intriguing evolutionary history and biogeography of Oreomyrrhis.
Cytological and morphological considerations
The base chromosome number for the genus Chaerophyllum is n = 11, with deviant and intraspecific variable counts only being reported in sect. Chaerophyllum (C. temulum [2n = 14 and 22] and C. procumbens [2n = 12 and 22]) (Pimenov et al., 2003
). Mapping base chromosome numbers onto the ITS phylogeny (Fig. 3) indicates that the numbers n = 6 and n = 7 occur only in the sect. Chaerophyllum clade (including Oreomyrrhis).
Although mericarps of most species of Oreomyrrhis are not equipped with structures such as hooks or nutrient tissues (Fig. 7) that could facilitate dispersal (Mathias and Constance, 1955
; van Steenis, 1963
), Raven (1973)
suggested that, during glacial maxima when high latitude alpine plants were forced to migrate to lower elevations closer to sea levels, chances for seed dispersals by transoceanic wanderers (e.g., albatross) would greatly increase. Compared to other species in Chaerophyllum, members of sect. Chaerophyllum (including Oreomyrrhis) possess relatively smaller mericarps (Spalik et al., 2001b
), with a further size reduction in Oreomyrrhis (Fig. 7). The evolution toward smaller mericarps in Oreomyrrhis might have been advantageous in enhancing their dispersibility (Howe and Smallwood, 1982
), especially when carried in mud adhering to the legs of wanderers (Salisbury, 1970
; Raven, 1973
).
|
; Baskin et al., 2004
). They differ from their Eurasian congeners by the absence of male flowers and by the possession of reduced corollas, traits also characterized by Oreomyrrhis and typically associated with self-pollination (Spalik and Downie, 2001
). Selfing was reported in the sub-Antarctic species O. hookeri (Moore, 1968
). Ongoing greenhouse bagging experiments indicate that the Australian species O. eriopoda is autogamous (K. Chung, unpublished data). In Oreomyrrhis, self-pollination in is also indicated by the very inconspicuous basal umbels of O. andicola, O. taiwaniana, and O. eriopoda, which are nearly sessile during early anthesis (K. Chung, personal observations). The same trait has also been noted in specimen labels of O. linearis (M. Coode & P. Stevens NGF 46182) and O. papuana (J. F. Veldkamp 6579). Despite inbreeding depression, the genetic benefits (e.g., reproductive assurance and purg
