| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Systematics |
2Department of Plant Biology, University of Minnesota, St. Paul, Minnesota 55108 USA; 3Brooklyn Botanic Garden, Brooklyn, New York 11225-1099 USA; 4Department of Plant Biology, Southern Illinois University, Carbondale, Illinois 62901 USA; 5Department of Biology, The Pennsylvania State University, University Park, Pennsylvania 16802 USA; 6Institut de recherche en biologie végétale, Université de Montréal, 4101 Sherbrooke East, Montréal, Quebec H1X 2B2 Canada; 7Bergius Foundation, Royal Swedish Academy of Sciences, 50017, SE-104 05, Stockholm, Sweden; 8Biology Department, Ithaca College, Ithaca, New York 14850 USA
Received for publication August 21, 2003. Accepted for publication January 22, 2004.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Key Words: Begonia • Begoniaceae • biogeography • divergence time • Hillebrandia • molecular phylogeny • paleoendemic • relict
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
; Swensen et al., 1998
; Soltis et al., 2000
; Schwarzbach and Ricklefs, 2000
; Wagstaff and Dawson, 2000
). These studies indicate that the Begoniaceae are most immediately related to a morphologically diverse clade that includes the Cucurbitaceae, Datiscaceae, and Tetramelaceae as well as three other families that were not traditionally linked with the begonias—the Anisophylleaceae, Coriariaceae, and Corynocarpaceae. These seven families constitute the order Cucurbitales (sensu APG II, 2003
), and recent investigations of wood anatomy in members of Cucurbitales have revealed some unusual features that are shared by all (Carlquist and Miller, 2001
). The Cucurbitales is noteworthy because its constituent families differ considerably in appearance from one another (but see Brouillet [2001]
for shared features and potential synapomorphies) and also because it includes families with large numbers of herbaceous species (Begoniaceae, Cucurbitaceae) as well as families with small numbers of predominantly woody species (Anisophylleaceae, Coriariaceae, Corynocarpaceae, and Tetramelaceae). Several genera in the order are either monotypic (Combretocarpus, Hillebrandia, Octomeles, Poga, and Tetrameles) or contain a small number of species (Corynocarpus–five species; Datisca–two species; Polygonanthus–two species). Recent molecular phylogenetic studies that have included members of the Begoniaceae (Swensen et al., 1998
; Schwarzbach and Ricklefs, 2000
; Wagstaff and Dawson, 2000
) all suggest that the Begoniaceae form a well-supported clade that is sister to the Datiscaceae sensu stricto.
Three genera are currently recognized within Begoniaceae: Begonia, Symbegonia, and Hillebrandia. Begonia is by far the largest of these, with roughly 1400 species. Most of the species are monoecious perennials, but a few differ by being either dioecious or annual. Begonia species typically inhabit moist, shady locations in humid lowland or upland forests, with the greatest number being found in middle-elevation and cloud forest habitats. Many of the species are narrow endemics. The species of Begonia are divided among 63 sections (Doorenbos et al., 1998
). Each section is restricted to a particular continent, with many sections and species occurring in Asia and America, relatively few in Africa, and none in Australia. The relationships among these sections are poorly understood, and a recent morphological study of the genus indicates that the circumscription of several sections is questionable (Doorenbos et al., 1998
). Recent efforts have focused on reconstructing Begonia phylogeny using DNA sequence data (Swensen et al., 2001
; Tebbitt et al., 2001
; Forrest and Hollingsworth, 2003
; Plana, 2003
). The genus Symbegonia includes 12 species that are endemic to New Guinea. It is traditionally distinguished from the other genera of Begoniaceae by having female flowers with a connate perianth and male flowers with free sepals and petals. Recent molecular-based phylogenetic analyses suggest that Symbegonia is nested within Begonia (Swensen et al., 1998
; Forrest and Hollingsworth, 2003
). The genus Hillebrandia is monotypic and represented by Hillebrandia sandwicensis, an endemic of the Hawaiian Islands and the focus of this study.
Hillebrandia sandwicensis was first described by Oliver (1866)
and named in honor of Dr. Wilhelm Hillebrand, a physician and botanist specializing in the Hawaiian flora. It is the only member of the Begoniaceae native to the Hawaiian archipelago. In appearance, Hillebrandia closely resembles Begonia but differs by its more numerous and more highly differentiated sepals and petals, its semi-inferior and incompletely closed ovary (inferior and completely closed in Begonia) (Gauthier, 1950
, 1959
; Charpentier et al., 1989
), fruits that dehisce between the styles (loculicidal or rarely septicidal or indehiscent in Begonia) and by the pattern of ornamentation of its pollen (van den Berg, 1983
). Hillebrandia flowers from February to June; after producing fruit, the aboveground parts die back to rhizomes, and the plants become dormant from late summer until January (MacCaughey, 1918
; Lorence, 1987
).
Hillebrandia sandwicensis is found on the islands of Kauai, Maui, and Molokai (Wagner et al., 1999
) and was reported once from Mount Ka'ala on Oahu (Hillebrand, 1888
) but is now thought to be extinct on that island. It is presently most abundant on Kauai and Maui (MacCaughey, 1918
), but even on these islands it is becoming increasingly rare (Lorence, 1987
). Throughout its range, the species is restricted to wet ravines in the montane rain forest zone at altitudes ranging from 900 to 1800 m (MacCaughey, 1918
), a habitat similar to that of many Begonia and Symbegonia species. Curiously, historical and current records of Hillebrandia populations show the species to be rare and localized despite the wider occurrence of suitable habitat, especially on Hawaii, the most recently formed of the Hawaiian Islands.
The seeds of Hillebrandia are very small (
0.4 mm in length) with some surface sculpturing. Similar seeds are found in Begonia (called "dust seeds") and can be effectively distributed by wind over long distances. However, the enclosed rainforest habitat of Hillebrandia (and Begonia) is not conducive to long-distance dispersal by wind. This situation presumably led Carlquist (1974)
to suggest that Hillebrandia seeds were most likely transported to the Hawaiian archipelago in mud adhering to the legs of birds, although anemochorous dispersal appears to be equally probable given the morphology of the seeds and a habitat little frequented by migratory birds. In Begonia, ITS-based phylogenies show clades that are strongly correlated with geographic origin, suggesting that long-distance colonization has occurred only rarely in the genus (Forrest and Hollingsworth, 2003
). Thus, while the habitat of Begoniaceae may prevent frequent anemochorous long-distance disperal events, these events appear to have occurred frequently enough to distribute Begonia around the globe and Hillebrandia onto the Hawaiian archipelago.
Previous phylogenetic analyses using chloroplast rbcL and nuclear 18S DNA sequence data support earlier morphologically based theories (e.g., MacCaughey, 1918
; Gauthier, 1950
, 1959
; Reitsma, 1983
) that Hillebrandia is the first branching member of the Begoniaceae (Swensen et al., 1998
). This is surprising given Hillebrandia's endemism to the Hawaiian archipelago, a very remote island chain whose current islands average only 15 million years in age coupled with the extremely large size and wide distribution of the genus Begonia.
This study aims to test previous assertions that Hillebrandia is the sister group to Begonia rather than embedded within it, to determine whether it is an ancient or a recent arrival on the Hawaiian Islands, and to propose a possible route by which it may have reached these islands. Because there is no fossil evidence available to assess the age of Hillebrandia directly, the minimal age of the Hillebrandia lineage is estimated here using analyses of sequence data calibrated by fossil dates suggested for close relatives.
An improved understanding of the phylogenetic placement and biogeography of Hillebrandia will facilitate a better understanding of evolutionary processes within the Begoniaceae as a whole.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
as modified by Rieseberg et al. (1992)
or using the DNeasy Plant DNA Extraction Kit (Qiagen, Valencia, California, USA). Other sequences were obtained directly from Genbank (see Appendix). Three gene regions were amplified for subsequent sequence analyses: rbcL, a highly conserved chloroplast gene encoding the large subunit of the protein ribulose bisphosphate carboxylase/oxygenase (rubisco); 18S rDNA, a highly conserved nuclear region encoding ribosomal RNA (rDNA); and ITS, the noncoding internal transcribed spacer regions adjacent to (and including) the 5.8S rDNA. Amplification primers for each gene region are listed in Table 1. The amplification parameters were specific to the gene of interest. For rbcL, an initial denaturation step at 95°C for 4 min was followed by 35 cycles of denaturation at 95°C for 1 min, annealing at 55°C for 1 min, extension at 72°C for 1 min, followed by a final extension step at 72°C for 4 min. For 18S, the same parameters were used as for rbcL, except that 30 cycles were run and the extension step was lengthened by 30 s. For ITS, the same parameters were used as for rbcL, except that the annealing temperature was 60°C. All amplification reactions were 50 µL in volume. Agarose gel electrophoresis followed by ethidium bromide staining was used to verify the presence and size of amplification products by comparison to a 1-kb ladder molecular mass standard (Life Technologies, Grand Island, New York, USA). Successful amplifications were purified using the QIAquick PCR Purification Kit (Qiagen, Valencia, California, USA) following the manufacturer's protocol. The products were eluted in 40 µL distilled water and again verified by agarose gel electrophoresis.
|
Sequence alignment and phylogenetic analysis
Two phylogenetic analyses were conducted: a taxonomically broad analysis of Cucurbitales using the rbcL and 18S sequences from 29 species and a more narrowly focused analysis of Begoniaceae using the rbcL, 18S, and ITS sequences from 31 species. Outgroups in both cases were chosen on the basis of previous analyses (Swensen, 1996
; Schwarzbach and Ricklefs, 2000
; Forrest and Hollingsworth, 2003
). In the Cucurbitales analyses, the outgroup consisted of two species belonging to the order Fagales (Myrica cerifera [Myricaceae] and Quercus rubra [Fagaceae]). In the Begoniaceae analyses, two Datisca species were designated as the outgroup. The Begonia species included in the Begoniaceae analysis were selected to represent the morphological and geographical ranges of the genus. All sequences included in these analyses are listed in the Appendix.
Alignment of the rbcL sequences was produced manually whereas 18S and ITS sequences were aligned using the Clustal X program (a graphical user interface for Clustal W; Thompson et al., 1994
), followed by manual editing. For ITS alignments, the gap opening penalty (GOP) and gap extension penalty (GEP) parameters were set to 10 and 5, respectively. For 18S alignments, the GOP and GEP parameters were 10 and 0.1 for pairwise alignments and 10 and 0.5 for multiple alignments.
Phylogenetic trees were estimated using maximum parsimony (MP), maximum likelihood (ML), and Bayesian inference. The MP analyses were conducted using PAUP* (Swofford, 2001
) via heuristic searches with tree bisection-reconnection (TBR) branch swapping and 1000 random-taxon-entry replications per search. Both combined and separate data sets were analyzed. A combinability test was not conducted for the combined data; recent studies have indicated that tests such as the ILD test (Farris et al., 1994
), if it can be run to completion, may find congruence despite the fact that different topologies are produced by the different data sets (Dowton and Austin, 2002
). Yoder et al. (2001)
showed an inverse relationship between data set congruence and phylogenetic accuracy. Dolphin et al. (2000)
also showed that phylogenetic noise can lead to significant incongruence. Bootstrap and decay analyses were conducted to assess branch support (Felsenstein, 1985
; Bremer, 1988
; Donoghue et al., 1992
). For bootstrap analyses, 100 bootstrapped data sets were analyzed using the same approach as with the original parsimony analyses. We used AutoDecay (Eriksson, 2002
) with PAUP* to calculate decay indices under parsimony. Constraint trees for each node were analyzed in PAUP* with converse constraints in effect. For each node, 100 random addition starting trees were swapped with TBR branch swapping.
For the maximum likelihood (ML) and Bayesian inference analyses, an evolutionary model was selected using MrModeltest 1.1b (J. Nylander, Uppsala University, personal communication), a simplified version of Modeltest 3.06 (Posada and Crandall, 1998
) that specifically tests the 24 models common to PAUP* and MrBayes (Huelsenbeck and Ronquist, 2001
). The hierarchical likelihood test in MrModeltest selected the most complex model available, the general time reversible model with gamma distribution of rates, and an estimated proportion of invariable sites (GTR + G + I). This model was selected for combined data sets as well as for the separate ones.
The ML trees were obtained in a two-step procedure. First, the parameters estimated by MrModeltest were used in PAUP*, which estimated a tree using 10 random addition sequence trees and TBR branch swapping. Parameters were then reestimated using the optimal tree found in step one, and this tree was swapped to completion.
The Bayesian inference analyses were performed using MrBayes 2.01 and the same model as in the ML analyses. In this case, all parameters were estimated during the analysis. Analyses started on random trees, and 1 000 000 generations of the Markov chain were run (mcmc) with four chains. A tree was sampled every 10 generations. For the Cucurbitales data set, 73 820 of the trees were used after the "burnin" of the chain was removed from consideration (generations prior to 261 800). For the Begonia data set, 92 200 of the 100 000 trees sampled were subsequently used to calculate clade posterior probabilities using a 50% majority-rule consensus tree (all trees sampled before generation 78 000 were excluded; the "burnin" of the chain). Data sets and trees resulting from these analyses can be obtained from http:// www.TreeBase.org.
Divergence time estimates
An estimate for the minimal age of Hillebrandia was calculated using two approaches: (1) a fixed-rate method and (2) two nonparametric methods that do not impose a fixed rate of nucleotide change for rbcL. All age estimates were obtained with the computer program r8s (Sanderson, 2002a
).
For the fixed-rate age estimation, the Langley-Fitch method (LF; Langley and Fitch, 1974
) was used to reconstruct divergence times using maximum likelihood under the assumption of a molecular clock. As part of this approach, a likelihood ratio test of rate constancy was conducted (Felsenstein, 1988
). For this analysis, we used only the best ML tree obtained from the combined Cucurbitales data set. This tree was rooted using the outgroups, saved with estimated branch lengths, and the outgroups were pruned prior to analyses in r8s. We used the parameters num_restarts = 5 and num_time_guesses = 5. Age constraints were enforced for three nodes: (1) Coriariaceae at minimum age 55 million years (my) BP (Saporta, 1865
), (2) Tetramelaceae at minimum age 55 my BP (Lankhanpal, 1970
), and (3) Cucurbitaceae set to a range from maximum 65 to minimum 39 my BP (Cronquist, 1981
; Muller, 1981
).
For the second approach, age estimates were obtained using nonparametric rate smoothing (NPRS; Sanderson, 1997
) and penalized likelihood (PL; Sanderson, 2002b
). Neither rely on a fixed evolutionary rate. We used the same starting tree and parameters as with the LF method. A cross-validation analysis was conducted for the PL analysis yielding a smoothing value of 562 (log 2.75).
To assess error levels in age estimates, we used a bootstrapping approach (Sanderson and Doyle, 2001
). One hundred bootstrap replicates of the data set were obtained using seqboot of the Phylip 3.5c package (Felsenstein, 1993
) and imported into PAUP* for branch length estimation. The tree was fixed for all replicates; branch lengths and all model parameters were estimated for each bootstrap replicate using the GTR + G + I model as before. Cross validation analyses were conducted in r8s for each replicate for use in the bootstrap PL analyses. Bootstrap log files were scanned and intervals of confidence (95%) for the estimated nodal ages were calculated using software available from one of the authors (T. Eriksson).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
|
Parsimony analysis of the combined rbcL, 18S, and ITS sequence data for the Begoniaceae resulted in a single most parsimonious tree (Fig. 2; length = 2561; CI = 0.4611; RI = 0.4407; RC = 0.2428). A total of 4179 nucleotide positions were analyzed (1428 from rbcL, 1813 from 18S, and 939 from ITS). Of these characters, 3267 were constant, 355 were variable but parsimony uninformative, and 557 were parsimony informative. The Begoniaceae appear as a well-supported monophyletic group (bootstrap 100%; decay index = 124) with Hillebrandia sister to Begonia plus Symbegonia. Within Begoniaceae, species groups correspond to continental locations, with the African and Malagasy species forming a clade with low support (bootstrap <50%; decay index = 2) that is sister to the rest. Ten of the 11 American species sampled form a weakly supported clade (bootstrap <50%; decay index = 1). An additional American species, B. herbacea (Brazil), is sister to the clade joining the American and Asian clades. The 10 Asian species sampled form a clade with low support (bootstrap <50%, decay index = 2). This clade includes a representative of Symbegonia, the New Guinea endemic, a result that supports the transfer of the species within genus Symbegonia to Begonia as a new section (Forrest and Hollingsworth, 2003
).
|
Maximum likelihood and Bayesian inference
The best ML tree for the Cucurbitales rbcL + 18S combined dataset is shown in Fig. 3 (ln likelihood –10683.58853) with posterior probabilities indicated for each node. This tree was also sampled during the Bayesian analysis (in generation 322 170). The ML tree is not among the six most parsimonious trees found for the same data, but is five steps longer in terms of equally weighted parsimony. Its likelihood score is 0.9–1.2% better than those of the most parsimonious trees. Two topological differences exist between the ML tree (Fig. 3) and the MP consensus (Fig. 1). Datisca species are sister to Begoniaceae in the MP tree, but Octomeles/Tetrameles are sister to Begoniaceae in the ML tree. Coriaria appears with Cucurbitaceae in the MP tree, but with Corynocarpus in the ML tree. In both trees, however, the position of Hillebrandia relative to the rest of the Begoniaceae is the same.
|
|
2 = 202.70; df = 27).
|
51–65 my BP; Eocene-Paleocene) and the Begonia crown group (26–42 my BP; Oligocene-Eocene). In fact, Hillebrandia, or an ancestor to both Hillebrandia and Begonia, may have existed prior to these dates. Regardless, all estimates for the age of the node where the Hillebrandia lineage splits off from the Begonia lineage predate the origin of the oldest above-water island of the Hawaiian archipelago (Kure atoll; 30 my BP) by at least 20 million years. |
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
, has since been recognized in several classification schemes (Cronquist, 1981
; Thorne, 1992
; Takhtajan, 1997
). Boesewinkel (1984)
also supported this association on the basis of ovule and seed structure, and Brouillet (2001)
supported it on the basis of the herbacous habit, cymose inflorescences (instead of racemes of cymes), and bifid stylodia. The relationships of these two families and Tetramelaceae is further supported by the small, numerous, operculate seeds with honeycombed testa and spherical pollen. The inclusion of Begoniaceae within Cucurbitales is supported by recent molecular phylogenetic work (see citations in APG II, 2003
), and a close relationship of the Begoniaceae with Cucurbitaceae and Datiscaceae was also recognized in serological work using seed extracts by Kolbe and John (1979)
. Begoniaceae and Datiscaeae, like the Cucurbitaceae, also produce cucurbitacins (Doskotch et al., 1969
; Doskotch and Hufford, 1970
; Kupchan et al., 1972
).
The phylogenetic analyses presented here, as well as previous morphological comparisons of Hillebrandia and Begonia (Gauthier, 1950
, 1959
; van den Berg, 1983
; Charpentier et al., 1989
; Brouillet, 2001
), provides substantial evidence to support Hillebrandia as the sister to the rest of Begoniaceae; this verification is a crucial step in understanding the biogeography of this species and its endemism to Hawaii.
Age estimates
The three methods utilized vary in their estimates of nodal ages. The NPRS estimates are the youngest, while the PL or the clock-based LF estimates are consistently older and quite close to each other. Sanderson (2002b)
noted that NPRS tends to "overfit" the data, and this might account for the difference in estimated ages in our analyses. Another difference between NPRS and the other methods is the use of the LOG penalty function in the optimality criterion, which was necessary to be able to use NPRS for our data. Slight differences in settings between the methods might explain why the NPRS estimates are considerably lower. The PL method relies on striking a balance between a situation where any branch can have any rate and the strict clock situation where all branches have the same rate. This is accomplished through a "cross validation" procedure, which yields a low "smoothing" value for clearly non-clocklike data and a high value for clocklike data (Sanderson, 2002b
). In our case, the smoothing value was intermediate and even though our data are not clocklike, the age estimates are almost identical to those reported by the LF method. So far, these methods are not very well known, and the effect of various parameter settings has not been investigated in detail. It is clear, however, that the choice of estimation method (LF, PL, or NPRS) does not affect the overall conclusion from the age estimates that the Hillebrandia lineage is considerably more ancient than the Hawaiian islands. The age of Hillebrandia itself, however, cannot be known based on these estimates.
Biogeography of Hillebrandia
Phylogenetic placement and divergence time estimates both suggest that the split forming the Hillebrandia lineage is a relatively ancient event in the history of Begoniaceae and that Hillebrandia is the sister group to Begonia. Moreover, given that Hillebrandia is both morphologically and molecularly divergent from Datisca, the direct ancestor of Begonia and Hillebrandia appears to be extinct. While its endemism to the Hawaiian Islands might typically suggest that Hillebrandia is a more recent mainland derivative, our data provides no indication of a mainland predecessor or of recent colonization. Furthermore, in contrast to many Hawaiian endemics (e.g., Munroidendron, Reynoldsia, Tetraplasandra [Araliaceae] Plunkett et al., 1997
; Argyroxiphium, Dubautia, Wilkesia [Asteraceae], Baldwin and Robichaux, 1995
; Bidens [Asteraceae], Ganders et al., 2000
; Hesperomannia [Asteraceae], Kim et al., 1998
; Clermontia [Campanulaceae], Lammers, 1995
; Cyanea [Campanulaceae], Givinish et al., 1995
; Vaccinium [Ericaceae], Powell and Kron, 2002
; Geranium [Geraniaceae], Pax et al., 1997
; Scaevola [Goodeniaceae], Howarth et al., 2003
; Haplostachys, Phyllostegia, Stenogyne [Lamiaceae], Lindqvist and Albert, 2002
; Gossypium [Malvaceae], Dejoode and Wendel, 1992
; Metrosideros [Myrtaceae], Wright et al., 2001
; Pittosporum [Pittosporaceae], Gemmill et al., 2001
; Rubus [Rosaceae], Howarth et al., 1997
; Viola [Violaceae], Ballard and Sytsma, 2000
), Hillebrandia is sister to the rest of its family and older than the current Hawaiian Islands. As such, it appears to be the only known example of a relict plant genus within the Hawaiian flora. Hillebrandia is of particular interest because several oceanic island taxa previously considered to be relictual based on their morphology have proven to be derived (suggesting recent colonization) based on phylogenetic analysis of molecular data. Prominent examples include the Hawaiian Hesperomannia (Kim et al., 1998
), the Juan Fernandez Dendroseris (Kim et al., 1996
), and the endemic Echium species of Macronesia (Böhle et al., 1996
). While Hillebrandia is highly unusual in being a Hawaiian relict, it has been suggested that oceanic islands make good refugia for relict taxa because the maritime climate provides environmental stability during prolonged periods of widespread drought or cold elsewhere (summarized by Moore, 1979
). We suggest that Hillebrandia, in part, owes its survival as a relict endemic genus on the Hawaiian Islands because these islands may have experienced a relatively stable maritime climate.
If Hillebrandia is a relict species, with Hawaii serving as a refuge, what is the origin of the ancestor of Hillebrandia? The present distribution pattern of Hillebrandia on the Hawaiian Islands suggests that it has been island hopping from older to younger islands in the same manner as documented in other Hawaiian groups (e.g., Cyanea, Givnish et al., 1994
; silversword alliance, Baldwin and Robichaux, 1995
; Drosophila, Desalle, 1995
). However, unlike the members of these groups, Hillebrandia has not undergone speciation in association with the colonization of new islands and exhibits no morphological discontinuities between populations even when such populations are on different islands (W. L. Clement, personal observations). In fact, preliminary genetic diversity studies indicate similar levels of diversity in populations on Kauai and Molokai, suggesting that the existing populations may be remnants of much larger populations (W. L. Clement, unpublished data).
Hillebrandia is found at 900–1800 m elevation on Kauai, Maui, and Molokai and was historically collected at similar elevations on Oahu, but it is absent from the largest and most recently formed island, Hawaii. Because Hawaii contains suitable montane rainforest habitat for Hillebrandia, its absence from this relatively young island (700 000 years old) may be due to the fact that this island's montane habitat has not been available long enough to allow Hillebrandia to colonize. The oldest islands no longer provide suitable montane habitat for Hillebrandia. Hillebrandia may, however, have colonized its present range by island hopping from older, now denuded, members of the Hawaiian archipelago, which would have provided suitable habitat since approximately 30 my BP. Given the absence of Hillebrandia from Hawaii, the youngest of the Hawaiian Islands, these island-hopping events were most likely rare. One might wonder if Hillebrandia was able to island hop from as far as the Emperor Sea Mounts that lie northward beyond the Hawaiian archipelago; however, the geological history of these islands reveals at least one obstacle in this pathway. After the formation of the last volcano (Koko) in the Southern Emperor Sea Mount Chain approximately 48 my BP, no other islands above 300 m existed until the formation of Kure approximately 30 my BP (Carson and Clague, 1995
). Thus, by the time Kure began forming, all other high islands had been submerged for approximately 18 my. Assuming relatively constant climate and vegetational zones on the islands during this time, the altitudinal constraints of Hillebrandia and the absence of high island habitats for a significant period of time would have prevented arrival by this route. On the other hand, climate fluctuations over the past 50 my may have allowed Hillebrandia to survive at lower elevations. While detailed climate information for the entire history of the Hawaiian islands is lacking, data based on the last one million years indicate recurrent fluctuations in sea level and temperature (Nullet et al., 1998
). During periods of cooler temperatures and lower sea levels, vegetation common at higher elevations occupied lower but more narrow zones. For example, fossil pollen records dating from the last glacial maximum on Oahu (about 20 000 years ago) suggest that at that time temperatures were 3°–5°C cooler than present at any given altitude (Hotchkiss and Juvik, 1999
). This temperature decline corresponds to similar cooling in Brazil (Stute et al., 1995
), the eastern Pacific (Pisias and Mix, 1997
; Pisias et al., 1997
), and New Guinea (Hope and Tulip, 1994
). Colonization of Hillebrandia (or its ancestor) somewhere along the Emperor Sea Mount Chain might suggest a boreal or boreotropical origin. Extant begonias are almost all tropical (B. grandis grows in temperate regions) and relatively little (2.6%) of the Hawaiian taxa are thought to have a boreal origin (Fosberg, 1948
). A boreotropical origin (Wolfe, 1975
) in Eurasia would be consistent with the presence of derived taxa in South America, and with the disjunct distribution of Hillebrandia's sister genus Datisca, with species in southwestern North America and southwestern Asia (Fig. 5). This scenario, however, would not explain why the African and Malagasy species appear to be basal within the phylogeny while the Asian species are derived (Forrest and Hollingsworth, 2003
; Plana, 2003
), unless the African species represent the earliest relicts of extinct Eurasian ancestors. Under the boreotropical hypothesis, the ancestor of Hillebrandia may have dispersed across Asia toward the Pacific, then island hopped down the Emperor Sea Mount. The ancestor would had to have become extinct in Eurasia, possibly due to the cooling, drying climate at the end of the Eocene, followed by recolonization in Asia to produce the current distributions of the phylogenetically derived Asian begonias. American begonias would have been derived from Africa via trans-Atlantic dispersal or from Eurasia across the Bering land bridge or the lower latitude North Atlantic land bridge.
|
A third scenario is that Hillebrandia colonized the Hawaiian Archipelago from tropical America. Under this scenario, the early Begonia species would have had to colonize Madagascar and Africa from the Americas, then become extinct in the Americas, and then later recolonize this region from Africa, a scenario similar in complexity to both the boreotopical and South Pacific-Malesian origins, but with less support given that only two of the more basal genera of Cucurbitales have species in South America (Fig. 5).
While it is impossible to be certain of Hillebrandia's origins, either a South Pacific–Malesian or a boreotropical origin appears plausible. A South Pacific–Malesian origin is consistent with current theory that the vast majority of the Hawaiian flora is of Malesian, Austral, or Pacific origin (Wagner et al., 1999
). A boreotropic origin is not necessarily incompatible with a species radiation from Malesia, if the boreotropical proto-Begonia was pushed to southern Asia from more northern latitudes. The present-day boreotropic taxa have floristic affinities to Indomalaya (Wolfe, 1975
).
Conclusions
Recent molecular phylogenies have led to a better understanding of the patterns of colonization of the Hawaiian Islands and the geographic sources of such colonists (e.g., DeJoode and Wendel, 1992
; Baldwin and Robichaux, 1995
; Givinish et al., 1995
; Lammers, 1995
; Howarth et al., 1997
; 2003
; Pax et al., 1997
; Plunkett et al., 1997
; Kim et al., 1998
; Ballard and Sytsma, 2000
; Ganders et al., 2000
; Gemmill et al., 2001
; Wright et al., 2001
; Lindqvist and Albert, 2002
; Powell and Kron, 2002
). In all these cases, the Hawaiian Islands were colonized by derived members of their families. The position of Hillebrandia as sister to, rather than derived from within, the rest of the Begoniaceae and its minimal age of 51–65 my BP (predating the current Hawaiian Islands) makes Hillebrandia the only known relict genus in the Hawaiian flora. We suggest that Hillebrandia has survived on the Hawaiian Archipelago while its source populations have become extinct because the Hawaiian Islands have provided a relatively stable climate and suitable montane habitat for at least 30 million years and during that time Hillebrandia has island hopped from older, now-denuded islands to younger more mountainous ones. However, Hillebrandia must at some time have colonized the Hawaiian Archipelago. The geographic origin of these source populations unfortunately remains obscure; however, we suggest a boreotropic or a Malesian– Pacific origin is most likely. Fossil material referable to the Begoniaceae would further improve upon these conclusions, in particular the age estimate and the exact origin of Hillebrandia.
|
FOOTNOTES |
|---|
9 E-mail: sswensen{at}ithaca.edu
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Baldwin B. G. R. H. Robichaux 1995 Historical biogeographical and ecology of the Hawaiian silversword alliance (Asteraceae): new molecular phylogenetic perspectives. In W. L. Wagner and V. A. Funk [eds.], Hawaiian biogeography: evolution on a hotspot archipelago, 257–287. Smithsonian Institution Press, Washington, D.C., USA
Ballard J. H. E. K. J. Sytsma 2000 Evolution and biogeography of the woody Hawaiian violets (Viola, Violaceae): arctic origins, herbaceous ancestory and bird dispersal. Evolution 54: 1521-1532[ISI][Medline]
Boesewinkel F. D. 1984 Ovule and seed structure in Datiscaceae. Acta Botanica Neerlandica 33: 419-429[ISI]
Böhle U.-R. H. H. Hilger W. F. Martin 1996 Island colonization and evolution of the insular woody habit in Echium L. (Boraginaceae). Proceedings of the National Academy of Sciences, USA 93: 11740-11745
Bremer K. 1988 The limits of amino acid sequence data in angiosperm phylogenetic reconstruction. Evolution 42: 795-803[CrossRef][ISI]
Brouillet L. 2001 Begoniaceae and the Cucurbitales: a morphological phylogenetic analysis. In Botany 2001: plants and people, abstracts, 94. Botanical Society of America, Albuquerque, New Mexico, USA
Bult C. M. Källersjö Y. Suh 1992 Amplification and sequencing of 16/18S rDNA from gel-purified total plant DNA. Plant Molecular Biology Reporter 10: 273-284
Bult C. E. A. Zimmer 1993 Nuclear ribosomal RNA sequences for inferring tribal relationships within Onagraceae. Systematic Botany 18: 48-63[CrossRef][ISI]
Carlquist S. 1974 Island biology. Columbia University Press, New York, New York, USA
Carlquist S. R. B. Miller 2001 Wood anatomy of Corynocarpaceae is consistent with Cucurbitalean placement. Systematic Botany 26: 54-65[ISI]
Carson H. L. D. A. Clague 1995 Geology and biogeography of the Hawaiian Islands. In W. L. Wagner and V. A. Funk [eds.], Hawaiian biogeography: evolution on a hotspot archipelago, 14–29. Smithsonian Institution Press, Washington, D.C., USA
Charpentier A. L. Brouillet D. Barabé 1989 Organogénèse de la fleur pistillée du Begonia dregei et de l'Hillebrandia sandwicensis (Begoniaceae). Canadian Journal of Botany 67: 3625-3639[ISI]
Chase M. W. et al 1993 Phylogenetics of seed plants: an analysis of nucleotide sequences from the plastid gene rbcL. Annals of the Missouri Botanical Garden 80: 528-580[CrossRef][ISI]
Cronquist A. 1981 An integrated system of classification of the flowering plants. Columbia University Press, New York, New York, USA
Cruzan M. B. M. L. Arnold S. E. Carney K. R. Wollenberg 1993 CpDNA inheritance in interspecific crosses and evolutionary inference in Louisiana irises. American Journal of Botany 80: 344-350[CrossRef][ISI]
Davidson C. 1973 An anatomical and morphological study of Datiscaeae. Aliso 8: 49-110
DeJoode D. R. J. F. Wendel 1992 Genetic diversity and origin of the Hawaiian Island cotton, Gossypium tomentosum. American Journal of Botany 79: 1311-1319[CrossRef][ISI]
Desalle R. 1995 Molecular approaches to biogeographic analysis of Hawaiian Drosophilidae. In W. L. Wagner and V. A. Funk [eds.], Hawaiian biogeography: evolution on a hotspot archipelago, 72–89. Smithsonian Institution Press, Washington, D.C., USA
Dolphin K. R. Belshaw C. D. L. Orme D. L. J. Quicke 2000 Noise and incongruence: interpreting the results of the ILD test. Molecular Phylogenetics and Evolution 17: 401-406[CrossRef][ISI][Medline]
Donoghue M. J. R. G. Olmstead J. F. Smith J. D. Palmer 1992 Phylogenetic relationships of Dipsacales based on rbcL sequences. Annals of the Missouri Botanical Garden 79: 333-345
Doorenbos J. M. S. M. Sosef J. J. F. E. De Wilde 1998 The sections of Begonia including descriptions, keys and species lists (Studies in Begoniaceae VI). Wageningen Agricultural University Papers 98: –2 266
Doskotch R. W. C. D. Hufford 1970 Hexanor–cucurbitacin D, a degraded cucurbitacin from Begonia tuberhybrida var. alba. Canadian Journal of Chemistry 48: 1787-1788[CrossRef][ISI]
Doskotch R. W. M. Y. Malik J. L. Beal 1969 Cucurbitacin B, the cytotoxic principle of Begonia tuberhybrida var. alba. Lloydia 32: 115-122[ISI][Medline]
Dowton M. A. D. Austin 2002 Increased congruence does not necessarily indicate increased phylogenetic accuracy—the behavior of the incongruence length difference test in mixed-model analyses. Systematic Biology 5: 19-31
Doyle J. J. J. L. Doyle 1987 A rapid isolation procedure for small quantities of fresh leaf tissue. Phytochemical Bulletin 19: 11-15
Eckenrode V. K. J. Arnold R. B. Meagher 1985 Comparison of the nucleotide sequences of soybean 18S rRNA with the sequences of other small-subunit rRNAs. Journal of Molecular Evolution 21: 259-269
Eriksson T. 2002 AutoDecay, version 5.0.3. Distributed by the author, Bergius Foundation, Royal Swedish Academy of Sciences, Stockholm, Sweden
Farris J. S. M. Kallersjö A. G. Kluge C. Bult 1994 Testing significance of incongruence. Cladistics 10: 315-319[CrossRef][ISI]
Felsenstein J. 1985 Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39: 783-791[CrossRef][ISI]
Felsenstein J. 1988 Phylogenies from molecular sequences: inference and reliability. Annual Journal of Genetics 22: 521-565[CrossRef]
Felsenstein J. 1993 PHYLIP (phylogeny inference package), version 3.5c. Distributed by the author. Department of Genetics, University of Washington, Seattle, Washington, USA
Forrest L. L. P. M. Hollingsworth 2003 A recircumscription of Begonia based on nuclear ribosomal sequences. Plant Systematics and Evolution 241: 193-211[CrossRef][ISI]
Fosberg F. R. 1948 Derivation of the flora of the Hawaiian Islands. In E. C. Zimmerman [ed.], Insects of Hawaii, vol. 1, 107–119. University of Hawaii Press, Honolulu, Hawaii, USA
Ganders F. R. M. Berbee M. Pirseyedi 2000 ITS base sequence phylogeny in Bidens (Asteraceae): evidence for the continental relatives of Hawaiian and Marquesan Bidens. Systematic Botany 25: 122-133[CrossRef][ISI]
Gauthier R. 1950 The nature of the inferior ovary in the genus Begonia. Contributions Institut de Botanique Universite de Montréal 66: 5-91
Gauthier R. 1959 L'anatomie vasculaire et l'interprétation de la fleur pistillée de l'Hillebrandia sandwicensis Oliv. Phytomorphology 9: 72-87
Gemmill C. E. C. G. J. Allan W. L. Wagner E. A. Zimmer 2001 Evolution of insular Pacific Pittosporum (Pittosporaceae): origin of the Hawaiian radiation. Molecular Phylogenetics and Evolution 22: 31-42[CrossRef][ISI]
Givnish T. J. K. J. Sytsma J. F. Smith W. J. Hahn 1994 Thorn-like prickles and heterophylly in Cyanea: adaptations to extinct avian browsers in Hawaii?. Proceedings of the National Academy of Sciences, USA 91: 2810-2814
Givnish T. J. K. J. Sytsma J. F. Smith W. J. Hahn 1995 Molecular evolution, adaptive radiation, and geographic speciation in Cyanea (Campanulaceae, Lobelioideae). In W. L. Wagner and V. A. Funk [eds.], Hawaiian biogeography: evolution on a hotspot archipelago, 288–337. Smithsonian Institution Press, Washington, D.C., USA
Hillebrand W. 1888 Flora of the Hawaiian Islands:. a description of their phanerogams and vascular cryptogams; annotated and published after the author's death by W. F. Hillebrand. B. Westermann, New York, New York, USA
Hiratsuka J. et al 1989 The complete sequence of the rice (Oryza sativa) chloroplast genome: intermolecular recombination between distinct tRNA genes accounts for a major plastid DNA inversion during the evolution of the cereals. Molecular and General Genetics 217: 185-194
Hope G. J. Tulip 1994 A long vegetation history from lowland Irian Jaya, Indonesia. Palaeogeography, Palaeoclimatology and Palaeoecology 109: 385-398
Hotchkiss S. J. O. Juvik 1999 A late Quaternary pollen record from Ka'au crater, O'ahu, Hawai'i. Quaternary Research 52: 115-128[CrossRef][ISI]
Howarth D. G. D. E. Gardner C. W. Morden 1997 Phylogeny of the Rubus subgenus Idaeobatus (Rosaceae) and its implications toward colonization of the Hawaiian Islands. Systematic Botany 22: 433-441[CrossRef][ISI]
Howarth D. G. M. H. G. Gustafsson D. A. Baum T. J. Motley 2003 Phylogenetics of the genus Scaevola (Goodeniaceae): implications for dispersal patterns across the Pacific Basin and colonization of the Hawaiian Islands. American Journal of Botany 90: 915-923
Huelsenbeck J. P. F. Ronquist 2001 MrBayes: a program for the Bayesian inference of phylogeny, version 2.01. Computer program distributed by the authors. University of California, San Diego, California, USA
Juncosa A. M. B. P. Tomlinson 1988 A historical and taxonomic synopsis of Rhizophoraceae and Anisophyllaceae. Annals of the Missouri Botanical Garden 75: 1278-1295[CrossRef][ISI]
Kim H.-G. S. C. Keeley P. S. Vroom R. K. Jansen 1998 Molecular evidence for an African origin of the Hawaiian endemic Hesperomannia (Asteraceae). Proceedings of the National Academy of Sciences, USA 95: 15440-15445
Kim S.-C. D. J. Crawford R. K. Jansen 1996 Phylogenetic relationships among the genera of the subtribe Sonchinae (Asteraceae): evidence from ITS sequences. Systematic Botany 21: 417-432
Kolbe P.-K. J. John 1979 Serologische Untersuchungen zur Systematik der Violales. Botanische Jahrbücher für Systematik Pflanzengeschichte und Pflanzengeographie 101: 3-15
Kupchan S. M. C. W. Sigel L. J. Guttman R. J. Restivo R. F. Bryan 1972 Datiscoside, a novel antileukemic cucurbitacin glycoside from Datisca glomerata. Journal of the American Chemical Society 94: 1353-1354[CrossRef][ISI][Medline]
Kuzoff R. K. J. A. Sweere D. E. Soltis P. S. Soltis E. A. Zimmer 1998 The phylogenetic potential of entire 26S rDNA sequences in plants. Molecular and Biological Evolution 15: 251-263
Lammers T. G. 1995 Patterns of speciation and biogeography in Clermontia (Campanulaceae, Lobelioideae). In W. L. Wagner and V. A. Funk [eds.], Hawaiian biogeography: evolution on a hotspot archipelago, 338–362. Smithsonian Institution Press, Washington, D.C., USA
