| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Systematics |
2Department of Botany, University of Wisconsin-Madison, 132 Birge Hall, Madison, Wisconsin 53706 USA; 3Georgetown College, 108 White-Gravenor, Georgetown University, Washington, D.C. 20057 USA; 4Department of Biological Science, Florida State University, Tallahassee, Florida 32306 USA
Received for publication June 17, 2003. Accepted for publication October 30, 2003.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Key Words: biogeography • Fuchsia • ITS • molecular clock • Onagraceae • penalized likelihood • rpl16 intron • trnL-F spacer
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
; Berry, 1982
; Godley and Berry, 1995
; Table 1). The genus consists mostly of mesic shrubs confined to cool, moist habitats, with close to three-quarters of the species concentrated in the tropical Andes. The remaining species occur in Mexico and Central America, Hispaniola, southeastern Brazil, and the southern Andes. In a notable disjunction, four species are native to the South Pacific, three on New Zealand, and one on the island of Tahiti.
|
suggested that the genus may have reached New Zealand from South America via long-distance dispersal across the Pacific, using Tahiti as a stepping stone. However, with the documentation of numerous fossil records of Fuchsia pollen from both Australia and New Zealand dating back to the Late Oligocene (Berry et al., 1990
), it appeared more likely that Fuchsia formed part of the Antarcto-Tertiary Geoflora that could have spread between South America, Antarctica, and Australia by either direct overland connections or across narrow water gaps during the Eocene and possibly into the Oligocene. In terms of vicariance biogeography, a significant geographic separation between the South Pacific and the neotropical species of Fuchsia would have been established at least 35–40 million years ago (mya). For the Tahitian species, a long-distance dispersal event from a New Zealand ancestor was hypothesized by Sytsma et al. (1991)
, based on chloroplast DNA restriction site analysis and the isolation and recent volcanic origin of Tahiti (±2 mya; Dymond, 1975
).
Among the American sections of Fuchsia, section Quelusia has a disjunct distribution between the southern temperate forests of the Andes and the cool montane forests of southeastern Brazil. This follows a pattern shared with other austral-Antarctic groups such as Araucaria, Cordyline, Drimys, and Griselinia, which inhabit the relicts of a once widespread Southern Hemisphere temperate forest (Zinmeister, 1987
; Berry, 1989
; Katinas et al., 1999
). Based on the evidence for austral biogeographical connections in Fuchsia during the Tertiary from section Quelusia and from the South Pacific species, along with suggestions that the more species-rich tropical Andean sections were more recent radiations associated with the Neogene uplift of the northern Andes (Berry, 1982
, 1985
), Berry (1989)
postulated that southern South America was the most likely area of origin for the genus.
The first molecular studies of Fuchsia used chloroplast restriction site mutations to explore the relationships of exemplars from each of the currently recognized 11 sections (Sytsma and Smith, 1988
; Sytsma et al., 1991
). The consensus cladogram of the combined studies (Sytsma and Smith, 1992
) showed the South Pacific species to be sister to the rest of the genus. The next branch was the monotypic section Jimenezia from Central America, followed by a trichotomy formed by section Quelusia along with the monotypic F. lycioides (section Kierschlegeria) from central Chile, the Andean section Hemsleyella, and a clade consisting of the remaining South and Central American sections. It was later realized (P. E. Berry, unpublished data) that the identity of the F. jimenezii accession in these studies was suspect and probably represented a cultivar of mixed parentage. In several family-wide molecular analyses of the Onagraceae, the circumboreal genus Circaea has consistently received strong support as the sister group of Fuchsia (Bult and Zimmer, 1993
; Conti et al., 1993
; Levin et al., 2003
). This is paradoxical with the proposed austral origin of Fuchsia suggested earlier.
The current study uses a more complete sampling of species within Fuchsia than did previous studies, as well as sequence data from both the nuclear and chloroplast genomes, to (1) determine if the morphologically based sectional delimitations are supported by the molecular data and (2) resolve basal relationships in the genus to determine where and when the initial diversifications occurred.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
; Conti et al., 1993
; Levin et al., 2003
). Within Fuchsia, 37 accessions were used, comprising 32 species, one natural hybrid, and one cultivar. At least one species from all 11 currently recognized sections in the genus are represented. Two different accessions were sampled for F. fulgens, F. triphylla, and F. boliviana. The 41 accessions with their corresponding voucher information and GenBank accessions are listed on the Botanical Society of America website (Appendix 1, in Supplemental Data accompanying the online version of this article). Most samples were taken from cultivated plants, which were generally progeny of well-documented, wild-collected plants. The exceptions include F. hybrida, a cultivar of mixed parentage mainly involving members of section Quelusia (Berry, 1989
). Also, the different accessions of both F. fulgens and F. triphylla were taken from plants that have long been cultivated. Both species are known to have been used in crosses to develop novel cultivars (Manthey, 1987
), so some of these accessions may have undergone crossing with other species or cultivars at some point in their history; morphologically, however, they are within the normal range of variation of wild-collected members of their species. Finally, the accession of F. vargasiana differs somewhat from wild-collected plants and may have been inadvertently crossed with another species or cultivar in its recent cultivation history.
Extractions, amplification, and sequencing
Total genomic DNA was extracted from fresh, frozen, silica-dried, or herbarium samples using a modified cetyl trimethyl ammonium bromide (CTAB) method (Doyle and Doyle, 1987
; Smith et al., 1991
) or DNeasy Plant Mini Kits (Qiagen Corporation, Valencia, California, USA). Standard polymerase chain reaction (PCR) and cycle sequencing techniques were used to amplify and sequence double-stranded DNA. The trnL-trnF spacer was amplified using primers "e" and "f" (Taberlet et al., 1991
), and the rpl16 intron was amplified using primers F71 (Jordan et al., 1996
) and R1516 (Kelchner and Clark, 1997
). Amplification of the internal transcribed spacer (ITS) region of nuclear ribosomal DNA, composed of ITS1, the 5.8S gene, and ITS2 (Baldwin, 1992
; Baldwin et al., 1995
) was conducted using primers ITS4 (5'-TCCTCCGCTTATTGATATGC- 3'; White et al., 1990
) and ITS5HP (5'-GGAAGGAGAAGTCGTAACAAGG-3'; Hershkovitz and Zimmer, 1996
). Additional sequencing primers used were ITS2 (5'-CGTAGCTACTTCTTGCATCG-3'; White et al., 1990
), ITS3B (5'-GCATCGATGAAGAACGTAGC-3'; White et al., 1990
), C5.8S (5'-TGCGTTCAAAGACTCGAT-3'; Suh et al., 1993
), and N5.8S (5'- ATCGAGTCTTTGAACGCA-3'; Suh et al., 1993
). The PCR products were cleaned using QiaQuick PCR purification kits (Qiagen) or the polyethylene glycol (PEG)/NaCl method of Kusakawa et al. (1990)
. Sequences were generated on an ABI Prism 373 or 377 DNA sequencer (Applied Biosystems, Foster City, California, USA).
Sequences were initially aligned manually using Sequencher 3.0 (Gene Codes Corporation, Ann Arbor, Michigan, USA). Insertions and/or deletions (indels) that were parsimony-informative (e.g., shared by two or more taxa) were recorded and mapped onto the trees. In the trnL-trnF spacer sequence of Fuchsia insignis, there was a large deletion (194 bp) in a region where other taxa had smaller indels. These smaller indels were scored when present, except for F. insignis, where they were treated as unknown. Areas of ambiguous alignment or containing poly-N stretches were excluded from the analyses.
Phylogenetic analysis
We reconstructed the phylogeny of Fuchsia and close relatives using both maximum parsimony (MP) and maximum likelihood (ML) optimality criteria in PAUP* version 4.0b8 (Swofford, 2002
). The heuristic MP searches employed 1000 random taxon addition sequences and tree bisection-reconnection (TBR) branch swapping. All characters were equally weighted and treated as unordered (Fitch, 1971
). Besides standard tree parameters (tree length, branch length [BL], consistency index [CI], and retention index [RI]), support for individual branches was estimated using nonparametric bootstrapping (BS; Felsenstein, 1985
) and decay analysis (index, DI; Bremer, 1988
). Bootstrap analyses used 1000 replicates (simple addition, saving up to 1000 trees per replicate, TBR branch swapping, Multrees) on the individual and combined data sets. Decay indices for branches were obtained using reverse topological constraint with 1000 random taxon addition sequences for each search, as suggested by Swofford (1993)
and implemented by Baum et al. (1994)
. Decay analyses were only performed on the combined chloroplast and nuclear data set.
Because an excessive number of most parsimonious trees did not allow searches to be completed for the combined cpDNA data set under the described search parameters, an initial heuristic search of 100 random addition replicates was conducted, with 10 trees saved per replicate. The resulting consensus tree was then used as a backbone constraint to search for trees inconsistent with the initial trees. This strategy searches for possible shorter trees and tests if the strict consensus tree reflects all most parsimonious trees, even though all equal-length trees have not been found (Catalán et al., 1997
).
The incongruence length difference (ILD) test (Farris et al., 1994
, 1995
) was employed to measure conflict between the three data sets, using the partition homogeneity test in PAUP*. We ran 1000 replicates on parsimony- informative characters using the TBR branch swapping algorithm (simple addition sequence, Multrees, steepest descent), with the number of trees retained for each replicate limited to 1000. This procedure may reduce the likelihood of finding most parsimonious topologies, but Farris et al. (1994)
noted that exact tree lengths are not critical to the test. Alternative relationships suggested by individual analyses were also examined by enforcing topological constraints while running maximum parsimony tests using 100 random addition replicates and calculating the number of additional steps required.
Maximum likelihood analyses were conducted on the individual and combined data sets in PAUP*. We used hierarchical likelihood ratio tests (hLRTs) in Modeltest 3.06 (Posada and Crandall, 1998
; alpha value of 0.01) to choose the preferred model of sequence evolution for the individual and combined data sets from among the 56 models tested by that program. An heuristic maximum likelihood search with TBR branch-swapping was then run using parameters determined by Modeltest for the selected model of sequence evolution.
Molecular clock analysis
A likelihood ratio test (Felsenstein, 1988
) was performed on the single ML tree (cpDNA and nuclear ITS sequences), by comparing the scores of the ML tree with and without the clock. This rejected the clock assumption, so we used the penalized likelihood (PL) method (Sanderson, 2002
) in the program r8s, version 1.60 (Sanderson, 2003
). The PL averages local differences in the rate of DNA evolution on different branches, taking into account the topology of branching. Unlike nonparametric rate smoothing (NPRS; Sanderson, 1997
), PL assigns a penalty for rapid rate changes among branches based on a smoothness parameter. We used the cross-verification approach in r8s to obtain the most likely smoothness parameter for this data. The ML tree was converted to an ultrametric tree (in which the lengths of all branches from the root are identical) in PL using the recommended algorithm = TN, collapsing zero length branches, and fixing the age of the Fuchsia-Circaea split. This split and thus calibration of the cpDNA and ITS sequence evolution for Fuchsia and relatives was based on results from a larger rbcL + ndhF analysis of Onagraceae and relatives in the Myrtales (Sytsma et al., in press
). Three fossils (Myrtaceae, Melastomataceae, and Combretaceae) generated quite similar rates for rbcL + ndhF sequence evolution and provided three time estimates for the Fuchsia–Circaea split (Table 2; see Sytsma et al., in press
, for details). These dates are consistent with the earliest known fossils of Circaea (fruits from the Oligocene in Europe; Boufford, 1982
) and Fuchsia (pollen from the Oligocene in Australia, Berry et al.; 1990
).
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
|
|
Our initial analyses were unable to run to completion due to an excessive number of most parsimonious trees, but of 520 trees recovered in our constraint analysis, the consensus tree had a length of 290, CI = 0.814, and RI = 0.828. We then ran a reverse constraint analysis, where PAUP* searched for any shortest trees incongruent with the strict consensus of the partial pool of shortest trees. During the constrained search, only trees one step longer (length = 291) were recovered, indicating that the initial consensus tree is representative of the most parsimonious trees. One of the most parsimonious trees is shown in Fig. 1, showing the branches that collapse in the consensus tree.
Modeltest 3.06 recognized the F82 + I + G model as the preferred one for the cpDNA data, with base frequencies A = 0.3776, C = 0.1710, G = 0.1670, T = 0.2843; proportion of invariable sites [I] = 0.5597; and gamma distribution shape [G] = 0.9157. An heuristic search resulted in one tree (ln L = –4316.72 555; tree not shown) that is nearly identical to the parsimony analyses—areas that collapse in the consensus tree are polytomies in the likelihood tree.
Analysis of ITS
The aligned length for the 41 accessions is 654 bp, with individual sequences ranging from 634 (F. cylindracea and F. triphylla) to 623 (Fuchsia brevilobis, F. glazioviana, and F. hatschbachii). A single indel groups all the Brazilian members sampled in section Quelusia. No nucleotide positions were excluded from analyses, and 123 (18.8%) characters were parsimony-informative. Two islands of trees were recovered (one island consisted of 18 trees and the other island of one) with length 312, CI = 0.776, and RI = 0.814. One of the most parsimonious trees is shown in Fig. 2, showing the branches that collapse in the consensus tree. With the increased number of parsimony-informative characters, there is higher support for some of the terminal relationships, but as with the cpDNA data, there is no clear support for basal relationships in the genus.
|
Combined analysis
Using a three-way comparison of the trnL-trnF, rpl16, and ITS data sets, the ILD test indicates significant incongruence (P = 0.001). Based on differences in the topologies of individual analyses, there were six taxa that had different positions in the cpDNA and ITS analyses (Fuchsia hybrida, F. magdalenae, F. pachyrrhiza, F. verrucosa, F. triphylla, and F. vargasiana). When an ILD test was conducted with these taxa removed, the data sets yielded insignificant P values, indicating congruence (P = 0.111). In most cases, the positions of the taxa that account for the nonhomogeneity between data sets vary within the same major clade in a way that does not contradict the overall results. In the ITS analyses, for example, F. magdalenae is sister to F. macrostigma, with these two taxa sister to F. andrei. In the cpDNA analyses, F. magdalenae is sister to F. nigricans and F. hartwegii. In both cases, F. madgalenae remains within the same moderately supported section Fuchsia clade. Fuchsia hybrida, which is a horticultural cultivar, groups with F. magellanica in the ITS analyses and with F. lycioides in the cpDNA analyses. In all analyses, F. lycioides is sister to F. magellanica, and this apparent conflict is consistent with F. hybrida being of mixed parentage with these two closely related species. Fuchsia pachyrrhiza is embedded within the section Fuchsia clade in the cpDNA analyses, while in the ITS analyses it is sister to the remaining members of the section Fuchsia clade; in both cases, there are no species from other sections embedded in the section Fuchsia + F. pachyrrhiza clade. In the ITS analyses, F. verrucosa is sister to section Hemsleyella, whereas in the cpDNA analyses it is sister to section Encliandra; these are conflicting results, but in both cases, F. verrucosa falls outside of section Fuchsia, where the species was previously placed. Lastly, the two accessions of F. triphylla are sister to section Hemsleyella in the cpDNA analyses but embedded within section Fuchsia in the ITS analyses. This could be due to past hybridization in cultivation within the triphylla cultivars. Because the alternative topologies of the taxa deemed incongruent by the ILD test largely do not involve shifts between major clades, we considered the incongruence to be minor and proceeded to combine the data sets.
Fitch parsimony analyses on the three combined data sets recovered one island of 156 trees (length = 626, CI = 0.764, RI = 0.600). Of the 2329 characters, 227 were parsimony- informative. One of the most parsimonious trees is shown in Fig. 3. As expected with increasing number of characters (Givnish and Sytsma, 1997
), the trees were more resolved than in the individual analyses. However, this increased resolution and support occurs mainly at the terminal branches. There was still very little resolution at the base of the tree.
|
Molecular clock analysis
Figure 5 shows the result of the PL analysis with a smoothing rate of 31.62. Using the intermediate B2 substitution rate for the Fuchsia data (Tables 2 and 3), the minimal age of the split between Hauya and Circaea occurs at about 52 mya (node A), and the Fuchsia-Circaea split occurred at least 41 mya (node B). According to the results of this analysis (all dates given are minimal ages), the initial diversification of what we now recognize as the main lineages of Fuchsia occurred about 31 mya (node C). The South Pacific Fuchsia lineage branches off around 30 mya, which is consistent with known fossil records of Fuchsia from Australia and New Zealand, which date back 25–30 mya (Berry et al., 1990
). The split between the Caribbean clade represented by F. triphylla and the much more species-rich Andean section Fuchsia occurred at about 25 mya (node D). It appears that at least an initial diversification of the large Andean section Fuchsia occurred over 22 mya (node E). The New Zealand section Procumbentes splits off from the other South Pacific section Skinnera at 18 mya (node F). Two other significant dates suggested by the molecular clock analysis are at nodes G and H. The first represents the split between the southern Andean species (F. lycioides and F. magellanica) and the Brazilian members of section Quelusia, at about 13 mya. The second, at just over 8 mya, represents the split between the New Zealand species of section Skinnera and the ancestor of the Tahitian F. cyrtandroides.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Within Fuchsia, the monophyly of most currently recognized sections is supported by the molecular data. For example, the six species sampled from section Quelusia all form a well-supported clade (BL = 6, BS = 87, DI = 3; Fig. 3), which in turn forms a more strongly supported clade with the monotypic sister section Kierschlegeria (BL = 11, BS = 99, DI = 7). Within section Quelusia, the southern Andean F. magellanica is supported as sister to the remaining, disjunct group of southern Brazilian species. The sole indel found in the ITS data set is a synapomorphy for the five Brazilian species sampled in section Quelusia (Fig. 2; Appendix 2, in Supplemental Data accompanying the online version of this article).
The South Pacific species are supported as a monophyletic clade, with the highly autapomorphic F. procumbens (section Procumbentes) sister to the members of section Skinnera, as previously found by Sytsma et al. (1991)
using chloroplast restriction enzyme analyses. There is similar strong support for section Ellobium (two of three species and three accessions sampled); section Encliandra (three of six species sampled); section Hemsleyella (three of 15 species sampled); and section Schufia (both species sampled); the latter is a close sister group to the monotypic section Jimenezia. For the large section Fuchsia (64 species), there is fairly weak support (BL = 3, BS = 66, DI = 1) grouping 10 of the 11 species sampled (excepting F. verrucosa). Because many of the informal species groups recognized by Berry (1982)
are represented in our sample, as well as anomalous or geographically outlying species (e.g., F. verrucosa, F. triphylla, and F. magdalenae), we do not expect the addition of other species from section Fuchsia to fall outside this group.
The single species excluded from section Fuchsia in the molecular analyses, F. verrucosa, is anomalous within that section, as pointed out by Berry (1982)
. Its large leaves and Andean distribution led to its original placement in section Fuchsia, but its small quadrangular flowers, very short floral tube, unique nectary type, large club-shaped style, long ovary, and pollen with smooth viscin threads are unlike any other member of the section (Fig. 6a). In the ITS analysis, F. verrucosa is sister to section Hemsleyella, but in the cpDNA and in the combined analysis, F. verrucosa is nested within a lineage of the Mexican and Central American sections Encliandra, Schufia, and Jimenezia. This is consistent with the small flower size, lobed nectaries, and smooth viscin pollen threads of F. verrucosa (Nowicke et al., 1984
). However, it is the only tetraploid species in this group (Berry, 1982
; Breedlove et al., 1982
), and while the axillary flowers are shared with section Encliandra, the large leaves are like those of sections Jimenezia and Schufia. Fuchsia verrucosa is morphologically so unlike any extant member of section Hemsleyella that the ITS results placing it sister to this group are puzzling, and it seems very unlikely this would be the result of some past hybridization event with a member of that section. Although the position of F. verrucosa is not well resolved among the Central American and Mexican sections of the genus, we are recognizing it (see below, Taxonomic changes) as a new section in the genus because of its disjunct geographical distribution in northern South America, its clear separation from section Fuchsia in the molecular analyses, and its possession of numerous autapomorphies. Interestingly, F. verrucosa shares a unique indel with F. jimenezii in the trnL-trnF spacer sequence. The possibility that these are sister species needs to be explored with additional molecular markers.
|
) and doubt about whether our accession of F. vargasiana (sister to pachyrrhiza in the combined analysis) had crossed with another species while in cultivation, it seems justifiable to maintain this as a distinct section from section Fuchsia. Also, F. pachyrrhiza lacks an indel in the rpl16 intron (Fig. 1; character 12 in Appendix 2, in Supplemental Data accompanying the online version of this article) that is present in all other Andean members sampled in section Fuchsia except for F. decussata. From the standpoint of our molecular results, F. pachyrrhiza is one of the best examples of the high degree of homoplasy in a number of key morphological characters that can occur in the genus. First, it has tubers as in the morphologically distinct sections Ellobium and Hemsleyella. It has alternate leaves, which occur elsewhere only in section Hemsleyella, the two South Pacific sections, and section Kierschlegeria. It has one of its whorls of stamens recurved into the floral tube, which only occurs elsewhere in sections Encliandra and Jimenezia (see Fig. 6b). Finally, its pollen has smooth viscin threads as in F. lycioides, F. verrucosa, and most of the Mesoamerican sections (Berry et al., 1988
).
Molecular clock and lack of basal resolution among sections
In contrast to the overall support for the current sectional delimitation of Fuchsia, we found little basal resolution in the genus. Prior hypotheses pointed to the South Pacific sections as being sister to the New World sections, with the austral sections Kierschlegeria and Quelusia basal within the New World lineages (Sytsma and Smith, 1988
; note that the basal position of section Jimenezia in that paper was likely due to a misidentification, particularly in light of the results obtained here). Although data from more genes would be desirable for our analysis, it is possible that the lack of basal resolution in Fuchsia is an indication of an ancient, rapid radiation, as was demonstrated by Fishbein et al. (2001)
in the Saxifragales. The PL analysis (Fig. 5) indicates that this initial radiation occurred about 31 mya, in the mid-Tertiary. It is unclear from our results where Fuchsia most likely originated, but given the strong support of the boreal genus Circaea as the closest outgroup of Fuchsia and of the Central American Hauya as the sister group to Circaea + Fuchsia, it seems just as likely that the genus arose in northern South America or even in southern North America or Mesoamerica, rather than in southern South America, as previously hypothesized (Berry, 1989
).
Our results confirm that there is a single South Pacific lineage (two sections, four species) in Fuchsia, as well as a separate southern South America/southern Brazil lineage (two sections, 10 species). The maximum likelihood tree also suggests that there was a single major Andean radiation, comprising sections Fuchsia, Hemsleyella, and Pachyrrhiza (a total of 80 species, including two on the Caribbean island of Hispaniola), as well as a separate appearance in South America of the monotypic section Verrucosa. Lastly, there was either a single evolutionary lineage that developed in Mexico and Central America, comprising modern-day sections Encliandra, Jimenezia, Schufia, and Ellobium, or alternatively two separate lineages developed there, one comprising section Ellobium (three species) and the second the other three sections (nine species).
Fossil pollen evidence indicates that Fuchsia was present in Australasia by the early Oligocene, about 30 mya, which was taken as evidence to support an early Tertiary origin of the genus in southern temperate forests (Berry et al., 1990
). The only extant sections in the genus that are found exclusively in the Southern Hemisphere are Skinnera and Procumbentes in the South Pacific and Kierschlegeria and Quelusia in South America. Given the long physical separation of Australia from Antarctica (ca. 50 mya) and the later opening of the Drake Passage between Antarctica and South America around 27 mya (Raven and Axelrod, 1974
), we expected to see sections Skinnera and Procumbentes appear as a well-supported sister clade to the American species, and within the New World species, sections Kierschlegeria and Quelusia as sister to the tropical Andean, Central American, and Caribbean taxa. It is important to note that our data do not contradict these relationships, but they offer no significant support for them either. The PL analysis places the South Pacific–American split minimally at about 30 mya, which is consistent with the fossil pollen data from Australasia, but somewhat late for a direct exchange between South America, Antarctica, and Australia (Taylor, 1991
, 1995
). According to our current understanding of Caribbean plate tectonics (Taylor, 1995
; Graham, 2003
), there could have been an opportunity for a fairly continuous biotic exchange between North and South America via the proto- Antilles in the current Panama–Costa Rica Arc during the mid- Tertiary, between 25 and 35 mya. This would be consistent with the diversification of separate South American and Central American lineages of Fuchsia at this time, according to our PL analysis.
Hummingbirds are the primary pollinators for all of the South American species of Fuchsia (Berry, 1982
, 1985
, 1989
). According to DNA hybridization distances among extant hummingbirds and a fossil divergence date corrected for incompleteness in the geological record, Bleiweiss et al. (1997)
and Bleiweiss (1998a)
determined that the major lineages of hummingbirds began to diverge in the Early Miocene (ca. 18 mya), even though the estimated divergence of hummingbirds from swifts dates back to the Paleocene (Sibley and Ahlquist, 1990
). Some hummingbird lineages, however, notably high-elevation Andean groups and North American taxa, have diverged only within the last 2–8 my, including the emeralds, the mountain gems, and the bee hummingbirds (Bleiweiss, 1998a
,b
). These radiations were presumably correlated with the marked uplift of the northern Andes in the Pliocene and Quaternary and the severe climatic fluctuations associated with a long sequence of alternating glacial and interglacial periods (van der Hammen, 1995
). For the Andean fuchsias, particularly the large section Fuchsia, we need a considerably denser molecular sampling of taxa to determine if certain groups, such as the high-elevation Fuchsia petiolaris species group (Berry, 1982
), are similarly the result of recent (e.g., Pliocene or Quaternary) radiations. According to Gentry (1982)
and Luteyn (2002)
, one reason the Neotropics are so much richer in plant species than the African or Asian tropics was the "explosive speciation" of many shrubby and herbaceous genera of plants centered in the northern Andes, including many hummingbird-pollinated groups like Fuchsia, Heliconia, Ericaceae, and Lobeliaceae.
There are only two Antillean species of Fuchsia, F. triphylla and F. pringsheimii, both native to Hispaniola. Our results indicate that F. triphylla was an early offshoot of section Fuchsia, possibly branching off from an Andean ancestor around 25 mya. Both F. triphylla and F. pringsheimii are tetraploid, whereas most members of section Fuchsia are diploid (Berry, 1982
). Although the two species form natural hybrids in areas of geographical overlap in the Dominican Republic (Berry, 1982
), they are morphologically very distinct. With wild-collected material of both species, we should be able to carry out molecular analyses and test if they are sister taxa and date their divergence.
The date obtained here for the split between the southern Andean and the southeastern Brazilian members of the Kierschlegeria + Quelusia clade, 13 mya, is significant because we could find no other molecular-based dates reported for this particular disjunction pattern. It will be particularly interesting to examine other clades with the same disjunction pattern, particularly more ancient lineages such as the southern Andean Araucaria araucana and the mainly Brazilian A. angustifolia, to determine if their divergence from a common ancestor in South America is concordant with the date obtained here for Fuchsia.
Finally, the date of 8 mya for the divergence of the Tahitian F. cyrtandroides from ancestors on New Zealand is close to the age of 10 mya that was estimated by Sytsma et al. (1991)
based on cpDNA restriction site mutations. Because Tahiti is part of a volcanic arc that emerged from the Pacific only 2–4 mya (Dymond, 1975
), this implies that F. cyrtandroides or its immediate ancestor either evolved for a considerable amount of time in New Zealand and then became extinct there after dispersing to Tahiti or else it dispersed earlier on to older, now submerged islands in the mid-Pacific.
Extensive morphological homoplasy
Our molecular results highlight a notable homoplasy in Fuchsia, namely the presence of male sterility, as manifested in species that are gynodioecious or dioecious (Fig. 4). In the subdioecious F. procumbens from New Zealand, the female is heterogametic, with male sterility controlled by a dominant gene (Godley, 1963
). The members of the closely related section Skinnera are gynodioecious, with the exception of F. cyrtandroides, which is hermaphroditic and also disjunct on the mid-Pacific island of Tahiti. The molecular results of Sytsma et al. (1991)
led to the conclusion that gynodioecy is the ancestral condition in the South Pacific sections, with F. cyrtandroides becoming secondarily hermaphroditic through the loss of male sterility (presumably from a founder population that was double recessive for the male sterility gene). In contrast, the male sterility that is known in all six species of section Encliandra is determined differently than in the South Pacific species; in this section, the hermaphrodites are heterogametic, and there is probably more than one gene involved (Breedlove, 1969
; Arroyo and Raven, 1975
). The lability of the loss or gain of male sterility in Fuchsia is also shown by the presence of some hermaphroditic populations of F. paniculata (section Schufia) north of the Isthmus of Tehuantepec in Mexico, but gynodioecious populations present in all populations south of the Isthmus and into Central America (Breedlove et al., 1982
). The closely related F. arborescens is entirely hermaphroditic. In South America, male sterility is known in F. lycioides (sole member of section Kierschlegeria; Atsatt and Rundel, 1982
), which is tetraploid and sister to a lineage of hermaphroditic species (section Quelusia). Recently, P. E. Berry (unpublished data) found the first evidence of male sterility in the large section Fuchsia, with populations of F. hartwegii from Colombia that are gynodioecious.
Fruits with few, large seeds characterize both sections Encliandra and Kierschlegeria, but these sections occur in different clades, making it likely that this character evolved independently in the two groups. Opposite or whorled leaves is the most common phyllotaxy in the genus, but alternate leaves characterize sections Kierschlegeria and Pachyrrhiza (both monotypic). In addition, section Hemsleyella is mostly alternate-leaved, but at least four of the 15 species in the section have opposite leaves. The two South Pacific sections are alternate-leaved as well, except for the opposite-leaved Tahitian F. cyrtandroides. This distribution of alternate leaves shows that this character is widely homoplasious in Fuchsia, even though most sections—including the large sections Fuchsia and Quelusia—have entirely opposite or whorled leaves.
Root tubers are of limited occurrence in Fuchsia, but they are found in most species of section Hemsleyella, in F. pachyrrhiza, and in some or all members of section Ellobium. None of these three groups show any indication of close sister group relationships in the molecular data, again pointing to separate origins of root tubers in the genus. Another seemingly apomorphic character in Fuchsia is the reflexion of the antepetalous set of anthers down into the floral tube, instead of the typical exserted arrangement (Figs. 4, 6b). This character occurs in sections Encliandra, Jimenezia, and Pachyrrhiza. The first two form part of a weakly supported Central American clade, but each is sister in turn to another section that has exserted anthers (Encliandra with Verrucosa and Jimenezia with Schufia). Axillary flowers are the norm for most sections of Fuchsia, but paniculate inflorescences characterize the Schufia + Jimenezia clade. In the large Andean sections Fuchsia and Hemsleyella, however, both axillary flowers and different types of terminal or axillary inflorescences are found, showing that inflorescences have evolved independently in several different lineages.
Earlier knowledge about extensive homoplasy in morphological characters in Fuchsia and the paucity of additional reliable characters had dissuaded us in the past from attempting a morphologically based cladistic analysis of the genus. In fact, most sections have been defined by unique combinations of character states, rather than by true synapomorphies. The results outlined earlier and shown in Fig. 4 support our earlier doubts about using morphological characters that in many other groups might be considered appropriate for cladistic morphological analysis. An analogous situation occurs in the Brassicaceae, which show many instances of convergence in morphological characters (e.g., gamosepaly, fruit morphology, and cotyledon position). Many of these characters were used in earlier classifications but have since been shown by molecular evidence to be poor predictors of phylogenetic relationships (Koch et al., 2003
).
Cytology and pollen morphology
Fuchsia and Circaea appear to share a basal chromosome number of n = 11 (Levin et al., 2003
; Fig. 4). Except for some members of the two largest sections of Fuchsia (sections Fuchsia and Hemsleyella), chromosome counts have been made for all remaining species and sections (Breedlove, 1969
; Berry, 1982
, 1985
, 1989
; Breedlove et al., 1982
; Berry et al., 1988
; Hoshino and Berry, 1988
, 1989
). Most species and sections of Fuchsia are consistently diploid (n = 11), with two-porate pollen. The entire Quelusia + Kierschlegeria clade, however, is polyploid (all are tetraploid except for some octoploid populations of F. regia), and all have three-porate pollen. There are no known cases of three-porate pollen in any diploid species of Fuchsia, but there are several cases of tetraploid species with entirely two-porate pollen, such as F. magdalenae, F. pringsheimii, and F. triphylla in section Fuchsia (Berry, 1982
) and F. verrucosa in its own new section (see Taxonomic changes, below). Two species in section Hemsleyella are tetraploid, and they have a mixture of two-porate and three-porate pollen (Berry, 1985
). If our interpretation of Fuchsia phylogeny is correct, then polyploidy evolved independently on at least four (but probably more) occasions in the genus: in the ancestor of sections Quelusia + Kierschlegeria (in association with three- porate pollen), in section Verrucosa, at least once in section Hemsleyella, and several times in section Fuchsia (Fig. 4).
Segmented-beaded viscin pollen threads characterize most species and sections of Fuchsia (Nowicke et al., 1984
), but the distribution of smooth viscin threads is mainly restricted to three Central American sections (Encliandra, Jimenezia, and Schufia) and the northern South American section Verrucosa, all of which form part of the same lineage in our maximum likelihood tree. Only two additional species, F. lycioides in the southern South American section Kierschlegeria and F. pachyrrhiza from Peru, also have smooth viscin threads (Nowicke et al., 1984
; Berry et al., 1988
).
Taxonomic changes
Fuchsia section Verrucosa P.E. Berry, sect. nov.—TYPE: Fuchsia verrucosa Hartweg (Fig. 6A). Frutex; folia opposita; flores axillares, bisexuales, ovario tetragono tubo florale duplo longiore, nectario cristis antepetalis; pollen biporatus filis rasilibus non segmentatis; bacca multiseminalis; chromosomatum numerus gameticus n = 22.
Shrubs; leaves opposite, elliptic, 5–16 x 2–8 cm; flowers solitary, axillary, bright orange; ovary elongate, tetragonous, 10–12 x 2.5–3 mm; floral tube four-angled, obconic, 3–6 mm long; petals 8–9 x 5–6 mm; nectary with four antesepalous ridged lobes 1.5 mm high; filaments erect, 2–3 and 1–1.5 mm long; style stout, 6–8 mm long, topped by a clavate stigma 2.5–4 mm long; pollen biporate, with smooth viscin threads; berry many-seeded; gametic chromosome number n = 22.
Conclusions
It is unclear from our molecular data how Fuchsia diverged early in its evolutionary history, but the data are consistent with an early, rapid divergence into several well- supported lineages that are now geographically and morphologically distinguishable, namely the South Pacific sections Skinnera and Procumbentes; the southern Andean/southern Brazilian sections Quelusia and Kierschlegeria; the Central American/Mexican sections Encliandra, Jimenezia, and Schufia together with the monotypic section Verrucosa of northern South America; the mainly tropical Andean sections Fuchsia, Hemsleyella, and Pachyrrhiza; and a possible second Central American lineage comprising section Ellobium. According to molecular clock analysis, Fuchsia diverged from a common ancestor with the boreal genus Circaea about 41 mya, and these two diverged from the Central American genus Hauya about 52 mya, which suggests that Fuchsia may not have had an austral origin as previously postulated. Similarly, the diversification of the tropical Andean fuchsias is considerably older than previously hypothesized (mid-Tertiary rather than late Tertiary or Quaternary), with the divergence of a small Antillean clade from South American ancestors occurring during the Miocene. The long evolutionary history of the genus may have contributed to the high levels of homoplasy evidenced in many key morphological characters, with multiple independent occurrences of polyploidy, male sterility, reflexed stamens, alternate leaves, and tuberous roots. Future molecular studies will address the origin of the major cultivar lines in Fuchsia to reconstruct which species and sections have been instrumental in the development of the myriad hybrids and cultivars. Also, carrying out a denser taxon sampling in the larger sections Fuchsia, Hemsleyella, Quelusia, and Encliandra will allow us to assess evolutionary and geographical relationships at considerably finer scales than in this paper.
|
FOOTNOTES |
|---|
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Atsatt P. R. P. W. Rundel 1982 Pollinator maintenance vs. fruit production: partitioned reproductive effort in subdioecious Fuchsia lycioides. Annals of the Missouri Botanical Garden 69: 199-208[CrossRef][ISI]
Baldwin B. G. 1992 Phylogenetic utility of the internal transcribed spacers of nuclear ribosomal DNA in plants: an example from the Compositae. Molecular Phylogenetics and Evolution 1: 3-16[CrossRef][Medline]
Baldwin B. G. M. J. Sanderson J. M. Porter M. F. Wojciechowski S. S. Campbell M. J. Donoghue 1995 The ITS region of nuclear ribosomal DNA: a valuable source of evidence on angiosperm phylogeny. Annals of the Missouri Botanical Garden 82: 247-277[CrossRef][ISI]
Baum D. A. K. J. Sytsma P. C. Hoch 1994 A phylogenetic analysis of Epilobium (Onagraceae) based on nuclear ribosomal DNA sequences. Systematic Botany 19: 363-388[CrossRef][ISI]
Berry P. E. 1982 The systematics and evolution of Fuchsia section Fuchsia (Onagraceae). Annals of the Missouri Botanical Garden 69: 1-198
Berry P. E. 1985 The systematics of the apetalous fuchsias of South America, Fuchsia section Hemsleyella (Onagraceae). Annals of the Missouri Botanical Garden 72: 213-251[CrossRef][ISI]
Berry P. E. 1989 A systematic revision of Fuchsia section Quelusia (Onagraceae). Annals of the Missouri Botanical Garden 76: 532-584[CrossRef][ISI]
Berry P. E. J. J. Skvarla A. D. Partridge M. K. Macphail 1990 Fuchsia pollen from the Tertiary of Australia. Australian Systematic Botany 3: 739-744[CrossRef]
Berry P. E. B. A. Stein S. Carlquist J. W. Nowicke 1988 Fuchsia pachyrrhiza, a new tuberous species and section of Fuchsia from northwestern Peru. Systematic Botany 13: 483-492
Bleiweiss R. 1998a Tempo and mode of hummingbird evolution. Biological Journal of the Linnean Society 65: 63-76[CrossRef]
Bleiweiss R. 1998b Origin of hummingbird faunas. Biological Journal of the Linnean Society 65: 77-97[CrossRef]
Bleiweiss R. J. A. W. Kirsch J. C. Matheus 1997 DNA hybridization evidence for the principal lineages of hummingbirds (Aves: Trochilidae). Molecular Biology and Evolution 14: 325-343[Abstract]
Boufford D. E. 1982 The systematics and evolution of Circaea (Onagraceae). Annals of the Missouri Botanical Garden 69: 804-994[CrossRef][ISI]
Breedlove D. E. 1969 The systematics of Fuchsia section Encliandra (Onagraceae). University of California Publications in Botany 53: 1-69
Breedlove D. E. P. E. Berry P. H. Raven 1982 The Mexican and Central American species of Fuchsia (Onagraceae). Annals of the Missouri Botanical Garden 69: 209-234[CrossRef][ISI]
Bremer K. 1988 The limits of amino acid sequence data in angiosperm phylogenetic reconstruction. Evolution 42: 795-803[CrossRef][ISI]
Bult C. J. E. A. Zimmer 1993 Nuclear ribosomal RNA sequences for inferring tribal relationships within Onagraceae. Systematic Botany 18: 48-63[CrossRef][ISI]
Catalán P. E. A. Kellogg R. G. Olmstead 1997 Phylogeny of Poaceae subfamily Pooideae based on chloroplast ndhF gene sequences. Molecular Phylogenetics and Evolution 8: 150-166[CrossRef][ISI][Medline]
Conti E. A. Fischbach K. J. Sytsma 1993 Tribal relationships in Onagraceae: implications from rbcL sequence data. American Journal of Botany 83: 221-233[CrossRef]
Doyle J. J. J. L. Doyle 1987 A rapid DNA isolation procedure for small quantities of fresh leaf tissues. Phytochemical Bulletin 19: 11-15
Dymond J. 1975 K-Ar ages of Tahiti and Moorea, Society Island, and implications for the hot-spot model. Geology 3: 236-240[Abstract]
Farris J. S. M. Källersjö A. G. Kluge C. Bult 1994 Testing significance of incongruence. Cladistics 10: 315-319[CrossRef][ISI]
Farris J. S. M. Källersjö A. G. Kluge C. Bult 1995 Constructing a significance test for incongruence. Systematic Biology 44: 570-572[CrossRef]
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 Review of Genetics 22: 521-565[CrossRef][ISI][Medline]
Fishbein M. C. Hibsch-Jetter D. E. Soltis L. Hufford 2001 Phylogeny of Saxifragales (Angiosperms, Eudicots): analysis of a rapid, ancient radiation. Systematic Biology 50: 817-847[CrossRef][ISI][Medline]
Fitch W. M. 1971 Toward defining the course of evolution: minimum change for a specific tree topology. Systematic Zoology 20: 406-416[CrossRef][ISI]
Gentry A. 1982 Neotropical floristic diversity: phytogeographical connections between Central and South America, Pleistocene climatic fluctuations, or an accident of the Andean orogeny?. Annals of the Missouri Botanical Garden 69: 557-593[CrossRef][ISI]
Givnish T. J. K. J. Sytsma 1997 Homoplasy in molecular vs. morphological data: the likelihood of correct phylogenetic inference. In T. J. Givnish and K. J. Sytsma [eds.], Molecular evolution and adaptive radiation, 55–101. Cambridge University Press, New York, New York, USA
Godley E. J. 1963 Breeding systems in New Zealand plants 2. Genetics of the sex forms in Fuchsia procumbens. New Zealand Journal of Botany 1: 48-52
Godley E. J. P. E. Berry 1995 The biology and systematics of Fuchsia in the South Pacific. Annals of the Missouri Botanical Garden 82: 473-516[CrossRef][ISI]
Graham A. 2003 Geohistory models and Cenozoic paleoenvironments of the Caribbean region. Systematic Botany 28: 378-386[ISI]
Hershkovitz M. A. E. A. Zimmer 1996 Conservation patterns in angiosperm rDNA ITS2 sequences. Nucleic Acids Research 24: 2857-2867
