| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
2 Botanical Institute, Göteborg University, Box 461, SE 405 30 Göteborg, Sweden
Received for publication December 1, 1998. Accepted for publication September 16, 1999.
ABSTRACT
Phylogenetic relationships of 38 species of the Alibertia group (Rubiaceae) and two outgroup species were investigated using the nuclear ribosomal 5S nontranscribed spacer (5S-NTS) and the internal transcribed spacers (ITS). Analysis of the data sets separately and in combination resulted in several well-supported and congruent groupings. However, the three analyses yielded different results as to the branching order of the basal clades. With the exception of Alibertia hispida, the species in the genus Alibertia appear in one weakly to moderately supported clade. This clade is in turn composed of two strongly supported subclades. One comprises several Alibertia species, including the type (A. edulis), three Borojoa species, and Randia tessmannii. The other subclade consists of Alibertia species only. This division is also generally supported morphologically by fruit size, corolla size, number of corolla lobes, and pollen aperture (porate vs. colporate). The sister group to the Alibertia clade comprises Duroia with Amaioua species internested. The close relationship of Ibetralia and Kutchubaea is corroborated. In addition, Alibertia hispida is a member of this strongly supported clade. Likewise, the two "Genipa" species are supported as a monophyletic group in 100% of the bootstrap replicates. It is concluded that the 5S spacer is superior to the commonly used ITS region in terms of resolution and robustness among closely related taxa.
Key Words: Alibertia • Gardenieae • ITS (internal transcribed spacer) • 5S-NTS (5S nontranscribed spacer) • molecular phylogeny • Rubiaceae
The informal Alibertia group (Rubiaceae-Gardenieae) as presently circumscribed (C. Persson, in press) includes ~11 genera with ~120 species. Their distribution is strictly neotropical, ranging from southern Mexico to northern Argentina (Andersson, 1992
). All members of the group are unarmed, dioecious shrubs or trees, with white or whitish flowers and contorted corolla aestivation. Like most genera in the Gardenieae, the many-seeded fruit is indehiscent and, with the exception of Amaioua, filled with a juicy pulp.
The Alibertia group was first proposed by Robbrecht and Puff (1986)
to accommodate six neotropical genera (Alibertia, Amaioua, Borojoa, Duroia, Genipa, and Kutchubaea), based on the shared possession of unisexual flowers. With the exception of Genipa and Borojoa the rest of the genera were included in Schumann's (1891)
subtribe Cordiereae (Borojoa was described in 1948).
Recently, Persson (1996; C. Persson, in press)
undertook two phylogenetic analyses of the Gardenieae, using morphological data and chloroplast DNA (cpDNA) sequence data from the rps16 intron and the trnL–F intergenic spacer. While both analyses suggest a similar circumscription of the Alibertia group, the group was more strongly supported by the cpDNA data than by the morphological data. Both analyses indicated that the type of Genipa (G. americana) should be excluded from the Alibertia group, a conclusion that was also reached in Andreasen and Bremer's analyses (1996, 1997)
of data from the rbcL gene. Furthermore, Persson's analyses indicated that the Alibertia group should be extended to include Melanopsidium from the Brazilian Atlantic forest, the fairly recently described genus Glossostipula (Lorence, 1986
), as well as the somewhat aberrant genus Stachyarrhena, whose male individuals have a spike-like inflorescence. Furthermore, the cpDNA analyses suggested that the Alibertia group should include three additional taxa that were not included in the morphological analysis. These were the monotypic genus Ibetralia, Genipa aff. williamsii, and one hitherto undescribed taxon from Colombia. Thus, the genus Genipa was shown to be paraphyletic. Both the morphological and the cpDNA analyses suggest that the Alibertia group only comprises neotropical, dioecious taxa with heteromerous flowers, and monad pollen grains. It is, therefore, thought that the recently described dioecious, monotypic genus Riodocea (Delprete, 1999
) with monad pollen from Espiritu Santo, Brazil, is also a member of the Alibertia group.
Although the Alibertia group was strongly supported by cpDNA data (C. Persson, in press), the underlying substitution rates in the rps16 and trnL–F sequences were too low for resolving relationships between closely related genera. Hence, only one clade was strongly supported by the bootstrap in all cpDNA analyses viz. the Alibertia–Borojoa clade. When the rps16 data were analyzed separately Ibetralia and Kutchubaea also formed a moderately supported clade. This clade also appeared as a moderately supported clade in the bootstrap analysis of the combined analysis, but did not appear among all shortest trees in the general combined cpDNA analysis. In the morphological analyses (Persson, 1996
), on the other hand, the internal branchings were fully resolved, but all clades were poorly supported. Furthermore, only the Alibertia–Borojoa and Amaioua–Duroia clades were retained after successive weighting. The close relationship between Alibertia and Borojoa has previously been hypothesized based on their close similarity of exotestal seed characters and pollen characters (Persson, 1993, 1995
). Several species of Alibertia, including the type A. edulis, and several Borojoa species have porate pollen grains, sometimes with a colpoid zone surrounding the pore, whereas other Alibertia species have distinctly colporate pollen. Likewise, A. edulis and Borojoa have secondary thickenings in the outer tangential wall of the exotesta, whereas these thickenings are absent in some of the other Alibertia species. These data indicate that Alibertia may well be paraphyletic if Borojoa is not included. Schultze-Motel (1986)
transferred several Borojoa species to Alibertia, but without stating the reason. The close relationship of Ibetralia and Kutchubaea in Persson's analysis (C. Persson, in press) is perhaps more surprising. When Bremekamp described the monotypic genus Ibetralia in 1934, he remarked that this species had certain characters in common with Alibertia, whereas it resembles Duroia and Amaioua in other characters. He did not mention Kutchubaea in this context. However, both Ibetralia and Kutchubaea differ from most genera in the Alibertia group in having more numerous corolla lobes (6–7 in Ibetralia and 6–11 in Kutchubaea). Of the remaining genera in the group, one might suspect that Amaioua and Duroia form a monophyletic group due to their unusual kind of stipules, which form a caducous circumscissile cap over the shoot apex. However, the relationships between the rest of the genera in the group are still obscure.
There has been considerable debate in the literature about whether or not different data sets should be analyzed separately or when combined. Some workers take the view that different data sets should always be combined (e.g., Donoghue and Sanderson, 1992
), whereas others argue for separate analyses of the data (e.g., Swofford, 1991
). Still others argue that the different data sets be subjected to a statistical test of homogeneity before deciding whether they should be combined or not (see Huelsenbeck, Bull, and Cunningham, 1996
, and references therein, for a review). In an explorative study of Solanaceae (Olmstead and Sweere, 1994
), for which different data sets were analyzed independently and in combination, it was concluded that reliance on any one of the analyses may result in lack of resolution, misleading interpretations, or both.
In this study I analyzed the internal transcribed spacer region (ITS) and the 5S nontranscribed region (5S-NTS) of the nuclear ribosomal DNA (nrDNA) separately and in combination for the Alibertia group. The ITS region has been widely used for inferring relationships at both infrafamilial as well as infrageneric level (see Baldwin et al., 1995
), whereas few studies have been performed using the 5S-NTS.
A major concern when dealing with ribosomal DNA is whether the sequences employed are orthologous or paralogous. Both the ITS sequences and the 5S sequences are arranged in one to several arrays of several hundred to several thousand tandemly repeated copies that occur at one or several chromosomal loci (Hillis and Dixon, 1991
; Playford, Appels, and Baum, 1992
; Baldwin et al., 1995
; Buckler, Ippolito, and Holtsford, 1997
). Copies at separate loci do not normally evolve independently, but in concert due to a number processes of which unequal crossing over and gene conversion seem to be the most important. Failure to homogenize copies of rDNA particularly occurs when the arrays of copies occur at different chromosomal loci (Hillis and Dixon, 1991
; Sastri et al., 1992
), but has also been demonstrated within individual arrays in the 5S rDNA in several plant groups (Cox, Bennett, and Dyer, 1992
; Kellogg and Appels, 1995
; Cronn et al., 1996
). This can in turn lead to sampling of sequences with different evolutionary histories (Doyle, 1992
; Baldwin et al., 1995
), which may differ from the organismic phylogeny.
The primary goals of the 5S-NTS and ITS analyses were (1) to investigate the relationships among genera of the Alibertia group and (2) to test the monophyly of the genus Alibertia. Additionally, this research presented the opportunity to compare directly the performance of ITS vs. 5S-NTS at this taxonomic level in Rubiaceae.
MATERIALS AND METHODS
Selection of taxa and sources of plant material
Thirty-eight ingroup species and two outgroup species, Randia aristeguietae and Rosenbergiodendron densiflorum, were sampled for nucleotide sequences of the ITS region and the 5S spacer. These taxa were chosen to include at least one representative of the 11 genera included in the Alibertia group (C. Persson, unpublished data). However, good and reliable sequences of the monotypic genus Melanopsidium were not obtained and accordingly were not included in the analyses. Moreover, I tried to analyze as many representatives as possible from the genera Alibertia and Borojoa. Details on plant material and vouchers are given in Table 1
.
|
|
, slightly modified with 1% PEG in the lysis buffer and an incubation temperature of 74°C. Following the purification in GENECLEAN® (BIO 101, Vista, California, USA) the DNA was resuspended in 10 mmol/L Tris (pH 8).
Amplification of genomic DNA was made on a Perkin Elmer (Norwalk, Connecticut, USA) Cetus DNA Thermal cycler 480 and a Perkin Elmer Gene Amp 9600. ITS 1 and ITS 2 were amplified separately mainly using the two external primers [5'-TAT GCT TAA AYT CAG CGG GT-3'] and [5'-AAC AAG GTT TCC GTA GGT GA-3'] and two internal ones [5'-GCT ACG TTC TTC ATC GAT GC-3'] and [5'-GCA TCG ATG AAG AAC GTA GC-3'] (Nickrent, Schuette, and Starr, 1994
). In addition, I used the universal primers (1–5) of White et al. (1990)
for a few samples.
For amplification of the 5S-NTS, I used the primers designed by Cox, Benett, and Dyer (1992)
: 5S forward [5'-TGG GAA GTC CTY GTG TTG CA-3'] and 5S reverse [5'-KTM GYG CTG GTA TGA TCG CA–3'].
Amplification of genomic DNA was either made in 50-µL reactions on a Perkin Elmer Cetus Thermal cycler 480 or in 25-µL reactions on a Perkin Elmer Gene Amp 9600. The 50-µL PCR (polymerase chain reaction) mixture contained 1.25 units of Taq polymerase (Advanced Biotechnologies, Epsom, UK), 0.2 µmol/L of each primer, 5 µL reaction buffer IV, 0.01 mg BSA (Pharmacia Biotech AB, Uppsala, Sweden), 10 mmol/L tetra methyl ammonium chloride (TMACl), 100 µmol/L of each dNTP, 2 µL of each DNA template, and 1 mmol/L MgCl2.
The ingredients of the 25-µL PCR reaction followed Struwe et al. (1998)
. Templates run on the Perkin Elmer Cetus Thermal cycler 480 were overlaid with mineral oil. For especially difficult samples, the MasterAmpTM PCR Optimization Kit (Epicentre Technologies corporation, Madison, Wisconsin) was employed according to the manufacturers instructions.
The cycle program included an initial incubation at 95°C for 1 min 50 s, followed by 30–40 (60) cycles of 50 s at 95°C, 50 s at 60°C, and 1 min 50 s at 72°C. Presence of fragments was checked on a Seakem® 1% agarose gel (FMC Bioproducts, Rockland, Maine, USA). About one-half of the products were run on a low melting gel, where the fragments were cut out and run a second time in the thermal cycler using the same cycling program as above.
Fragments were either purified with QIAquick spin columns (QIAGENTMGmbH, Hilden, Germany) or BIO 101's GENECLEAN® kit and subsequently separated on an ALFexpressTM (Pharmacia Biotech AB) machine or an ABI PrismTM 377 DNA Sequencer (Perkin Elmer Applied Biosystems, Foster City, California). The cycle sequencing for the ALFexpress was performed with Thermo sequenase® (Amersham fluorescent labeled primer cycle sequencing kit with 7-deaza-dGTP). A dRhodamine Terminator Cycle Sequencing Ready Reaction kit (Perkin-Elmer Applied Biosystems Division) with AmpliTaq DNA Polymerase, FS, was used for the ABI 377. The program for thermo-sequenase reactions comprised an initial incubation time at 96°C for 2 min followed by 18 cycles of 30 s at 95°C and 40 s at 60°C. The cycles ended with a 5-min extension at 60°C. The cycle sequencing program for the ABI 377 had 35 cycles of 15 s at 94°C, 15 s at 50°C, and 72°C for 4 min. Sequencing reactions for the ABI 377 were cleaned using Sephadex® G-50 Fine DNA Grade (Pharmacia Biotech AB).
Cycle sequencing fragments were separated either on a 0.5 mm Long Ranger Hydrolink gel (FMC Bioproducts, Rockland, Maine) or a Page-plus gel (AMRESCO® Solon, Ohio, USA). Raw data files were analyzed either with ABI PrismTM 377 Collection 2.1 package (Perkin Elmer Applied Biosystems) or AM V3.0 (Pharmacia Biotech AB).
The aligned matrices can be found at the BSA website ( http://www.botany.org/bsa/ajbsupp/v87/persson_5s.txt for the 5s file and http://www.botany.org/bsa/ajbsupp/v87/persson_its.txt for the ITS file).
Alignment and gap coding
Alignment and gap coding were done by hand in SequencherTM version 3.0 and are largely based on the principles outlined by Oxelman, Lidén, and Berglund (1997)
. Gaps were placed so as to keep the number of informative characters (indels and substitutions) to a minimum within an aligned sequence. When indels of equal length occur in more than one sequence they were considered homologous and coded as the same event if they could not be interpreted as a different insertion/deletion event. Indels were scored as separate characters and added to the sequence data matrix as additional binary characters. In regions comprising polyT tracts of unequal length, the indels were interpreted as duplications of adjacent sequences and have been coded as unordered multistate characters. Regions where alternative alignments were possible were excluded from the analyses.
Cladistic analyses
The aligned ITS and 5S spacer sequences were separately and jointly analyzed in PAUP version 3.1.1 (Swofford, 1993
) using maximum parsimony on a Macintosh PowerBook G3. The minimal length trees were searched for using the heuristic option with tree bisection reconnection (TBR) swapping, MULPARS off, and 500 replicates of random addition sequence order in effect. The resulting minimum length trees were used as starting trees in a new search using TBR, MAXTREES 2000, and MULPARS on.
Internal clade support was evaluated using the bootstrap procedure with 1000 replicates, no branch swapping, and five random entry orders per replicate.
RESULTS
The ITS analysis
The boundaries of the ITS region were determined by comparison to previous analyses (Baldwin, 1993
; Andreasen, Baldwin, and Bremer, 1997
). The total length of the ITS 1 and ITS 2 spacers ranged from 420 base pairs (bp) in Rosenbergiodendron to 483 bp in Alibertia elliptica. Duroia aquatica and D. eriopila had the shortest total length in the ingroup (429 bp). ITS 1 varied between 231 bp in Rosenbergiodendron and 265 bp in Alibertia elliptica, whereas Duroia aquatica, D. eriopila, and Amaioua guianensis showed the shortest length (235 bp) in the ingroup. ITS 2 ranged from 189 bp in Rosenbergiodendron to 224 bp in the undescribed Colombian species. The shortest ITS 2 sequence among the ingroup species was 194 bp in Duroia aquatica and D. eriopila. The size of the ITS sequences in the present study is within the range reported for other angiosperms (Baldwin et al., 1995
).
The total GC content of ITS 1 and ITS 2 is >70% in most species. The GC content of ITS 1 varies from 67% in Alibertia sessilis and A. concolor to 76% in Glossostipula concinna. The latter species also has the highest percentage (75%) in ITS 2, whereas the undetermined Borojoa (B. sp.) shows the lowest GC value in ITS 2 (64%). The GC content of the ITS loci is one of the highest among the angiosperms (cf. Baldwin et al., 1995
) and was probably the reason for many difficulties in obtaining PCR products when using a standard protocol. However, when employing the MasterAmpTM PCR Optimization Kit, an almost 100% recovery was obtained.
After exclusion of six regions where alternative alignments were possible, the aligned ITS sequences resulted in a 543-bp long data matrix, including gap codings. Unfortunately, no ITS 1 PCR product was obtained from Randia tessmannii. Hence, this sequence is lacking in the data matrix. A total of 45 gaps had to be inserted, of which 27 were informative. Seven of these were assigned to polyT tracts, whereas three were interpreted as duplication events.
The entire data matrix contained 132 (24%) phylogenetically informative characters (121 when excluding outgroup genera). Approximately 22% of the informative characters in the ITS spacer region are due to indels. Thus, the ITS region has mainly evolved by point mutations. However, the indel proportion is probably underestimated and would likely increase if it were possible to analyze the six excluded regions.
Cladistic analysis of the ITS region (ITS 1 and ITS 2) hit the maxtree value of 2000 most parsimonious trees. These were 542 steps with a consistency index (CI) of 0.71 (0.58 without uninformative characters) and a retention index (RI) value of 0.78. Bootstrap values are plotted on the branches in the strict consensus tree (Fig. 1).
|
The 5S analysis
As the exact boundaries of the 5S spacer could not be unambiguously determined, the aligned sequences were cut at an arbitrarily chosen position at both ends of the spacer where most sequences were readable.
As a result of failure to sequence with the cy5-labeled forward primer, parts of the 5S sequences are absent in 12 species. However, the length of the complete 5S sequences among the ingroup ranged from 371 bp in Alibertia edulis to 630 bp in A. pilosa and A. steinbachii. The outgroup species Rosenbergiodendron densiflorum falls below this range with a length of 334 bp.
The GC content of the 5S spacer varied from 30% in Alibertia elliptica to 51% in A. acuminata. Having aligned the 5S sequences, the matrix comprised 793 positions including gap codings. A total of 47 gaps had to be inserted, of which 21 were informative. Six of these represent polyT tracts and one is a duplication. Of the 793 positions, 267 (34%) were informative. The PAUP search yielded 1206 trees of 1015 steps. These had a CI of 0.78 (0.67 excluding uninformative characters) and an RI of 0.81. The strict consensus tree and the bootstrap values of individual clades are shown in Fig. 2.
|
Combined ITS and 5S-NTS analysis
The ITS data in conjunction with the 5S data yielded 54 most parsimonious trees of 1579 steps with a CI of 0.74 (0.63 without uninformative characters) and an RI of 0.79. The strict consensus tree and bootstrap values for the combined analysis are shown in Fig. 3.
|
DISCUSSION
This study suggests, for the Alibertia group at least, that the 5S nontranscribed spacer is superior to the ITS region in several respects. The low GC content, as compared to the very high GC content in the ITS sequences, makes 5S-NTS relatively easy to amplify. However, high sequence variability at the 5' end of the 5S spacer may cause problems for the primer to anneal. Multiple sequence alignment of the 5S spacer is relatively straightforward, and informative positions are fairly uniformly spread along the length of the spacer. Hence, in contrast to ITS, no entire regions were excluded from the analyses, only occasional very aberrant motifs. Moreover, the 5S-NTS provided approximately twice as many informative characters as the ITS sequences (267 vs. 132). In addition, the 5S spacer is more powerful than the frequently used ITS region for resolving clades at the infrageneric level in the Alibertia group, and the support for the resolved internal clades is strong in many cases. Likewise, the CI and the RI are higher in the 5S-NTS analysis.
As may be expected, the support for the strongly supported clades of the individual analyses increased when the two data sets were combined. On the other hand, most of the incongruent taxa (although weakly supported) in the individual analyses appear in the combined analysis as single branches on a basal polytomy. Furthermore, the CI value (excluding autapomorphies) is intermediate between those of the separate ITS and 5S values, whereas the RI is lower than the 5S value, but the same as the ITS value. On the other hand, the number of most parsimonious trees decreased considerably in the combined analysis.
Although the occurrence of divergent paralogues are probably ubiquitous among angiosperms (Buckler IV, Ippolito, and Holtsford, 1997
), no indications of paralogous loci were observed in the present analysis. No evidence of 5S nor ITS length variants were identified. The PCR products were present on the agarose gel as sharply delimited single bands. Furthermore, few individual positions showed double peaks (as determined by automated sequencing) that could indicate occurrence of paralogous loci.
Phylogenetic inferences
The three topologies from the individual and combined analyses of the ITS and the 5S regions are to a large extent congruent, having only a few, weakly supported, conflicting nodes. The taxa may be roughly divided into three major lineages, and a basal grouping where the different data sets yield alternative hypotheses (Figs. 1–3). The three major lineages (the Alibertia sessilis, Alibertia edulis, and Duroia-Amaioua clades) are strongly supported by bootstrap analyses, whereas most basal clades are weakly supported and inconsistent in their placement.
Basal branchings
Among the basal branchings, the clade comprising Genipa aff. williamsii and G. curviflora and the clade comprising Kutchubaea semisericea, Alibertia hispida, and Ibetralia surinamensis are the only ones that are strongly supported and appear in all three analyses. However, like the rest of the genera occurring basally, the placement of these clades varies between the different analyses.
Chloroplast DNA data suggested that the genus Genipa is paraphyletic (C. Persson, in press), the type species of Genipa americana being grouped together with paleotropical genera, whereas Genipa aff. williamsii is nested within the Alibertia group. Genipa aff. williamsii and G. curviflora can be morphologically distinguished from G. americana by their persistent, terminal, membraneceous stipules that hide the lower parts of the corolla tubes in the inflorescence, and by their four- to five-porate pollen grains. In addition, the reduced calyx is a striking feature of these species.
The close relationship between Ibetralia and Kutchubaea was also estimated by cpDNA data, whereas the placement of Alibertia hispida together with these genera is new. Similar to Ibetralia and Kutchubaea, A. hispida has a relatively large number of corolla lobes.
Like these two clades, Stachyarrhena, Glossostipula, and the undescribed Colombian species appear in different positions in every one of the three analyses, and in all analyses the support for their positions is poor.
The Duroia–Amaioua clade
The Duroia–Amaioua clade is strongly supported in the separate analyses as well as in the combined analysis. The ITS data suggest a weakly supported sister-group relationship with the Alibertia clades (including Borojoa) and a clade comprising Glossostipula, Stachyarrhena, and Alibertia bertierifolia. The combined and the 5S-NTS analyses also support a sister-group relationship with the Alibertia clades, however Glossostipula, Stachyarrhena and Alibertia bertierifolia are placed elsewhere. Morphologically, Duroia and Amaioua are easily recognized by their stipules, which form a circumscissile cap over the shoot apex. Traditionally, these two genera have been distinguished by the number of flowers in the female inflorescence, one to three in Duroia vs. several to many in Amaioua. However, the present study does not support such a subdivision. All three analyses suggest that Duroia is paraphyletic and that Amaioua is either a monophyletic group within Duroia, or possibly polyphyletic. The only strongly supported group in the bootstrap analyses of the 5S-NTS and the combined analyses comprises Duroia aquatica, D. eriopila, and D. micrantha. The two former species also appear as a moderately supported clade in the ITS bootstrap analysis. Duroia hirsuta is the sister to the remainder of the group in the ITS analysis and the combined analysis, whereas it appears with Amaioua guianensis as the sister to the rest in the 5S analysis. It differs from many other Duroia species in having threadlike calyx lobes, a feature that also occurs in many Amaioua species. In addition, Duroia hirsuta differs from most Duroia species in distally swollen shoots that are inhabited by ants. Karsten (1859)
accommodated D. hirsuta in a genus of its own, Schachtia.
The Alibertia edulis clade
The Alibertia edulis clade is strongly supported in all analyses and is composed of six Alibertia species, the three Borojoa species, and Randia tessmannii. The 5S and the combined analyses also strongly suggest that Alibertia bertierifolia should be included in the Alibertia edulis clade. The sister group to the Alibertia edulis clade is invariably the clade comprising the rest of the Alibertia species (the A. sessilis clade). The Alibertia edulis clade is largely a group of rainforest species that has its center of diversity in the western Amazon. The widespread species Alibertia edulis, however, occurs also in savannas. The species of the Alibertia edulis clade are recognized morphologically by a relatively large number of corolla lobes (5–6 in male flowers and 6–7, rarely 8 in female flowers), and by their relatively large fruits. All species are also characterized by porate pollen grains, and many species have seeds with thickened tangential walls of the exotesta cells.
The species of Borojoa considered here do not form a monophyletic group in any of the analyses, suggesting that this genus is unnatural. When Cuatrecasas (1948)
described the genus Borojoa on the basis of B. patinoi he considered that this genus is close to Einsteinia (= Kutchubaea), Kutchubaea and Duroia. In 1953, Cuatrecasas transferred several species from Thieleodoxa and Alibertia to Borojoa. He distinguished Borojoa from Alibertia by its larger fleshy fruits, at maturity completely mucilaginous with 6–8 carpels vs. smaller, leathery fruits with 3–5 carpels, and also by the misconception that the flowers were homomeric (i.e., the same number of corolla lobes in male and female flowers) in Alibertia.
Although the support for the entire group is strong, many of the internal clades are unresolved or have low support. In the 5S and the combined analyses the internal clades are essentially the same. In both analyses Alibertia edulis and A. acuminata are members of the same clade, and both share a 250-bp long deletion in the 5S-NTS. Another strongly supported clade consists of Alibertia latifolia, A. tutumilla, Borojoa sp., and R. tessmannii. When Standley described R. tessmannii, he remarked that this species does not agree well with any of the genera in the Gardenieae, but that it fits better in Randia than elsewhere. It is now generally agreed that the genus Randia should only comprise species with pollen grains in tetrads, and that most species are equipped with short shoots and thorns (C. Gustafsson, personal communication). Since R. tessmannii does not have any of these features, it was strongly suspected that it belonged in the Alibertia group.
The Alibertia sessilis clade
Like its sister group, the A. sessilis clade is strongly supported in all three analyses. Nine of the species in this group share an ~90-bp long insertion in the 5S-NTS sequence. The clade comprises only species that have traditionally been placed in Alibertia. Most of these species occur outside rainforest areas, and the group has its center of diversity in the somewhat drier states of eastern and southeastern Brazil. However, Alibertia garapatica and A. myrciifolia have somewhat deviant distributions. The first species ranges from southern Costa Rica to northern Colombia, whereas the latter occurs from Venezuela to Minas Gerais in Brazil. Furthermore, Alibertia pilosa occurs along the wet eastern slopes of the Andes between 100 and 1200 m. Most of the species in the A. sessilis clade are small shrubs, but Alibertia sessilis and A. macrophylla are small trees. In comparison with the species of the Alibertia–Borojoa clade, the corolla in the A. sessilis clade is normally smaller and the lobes are usually fewer (five or less in females and four or less in males). One exception is Alibertia pilosa in which female individuals have six (seven) corolla lobes, whereas the males have five corolla lobes. In addition, all species have colporate pollen grains. Few subordinate clades were resolved in the strict consensus tree using ITS data. However, Alibertia pilosa and A. steinbachii form a clade that is supported in 100% of the bootstrap replicates. Both these species have a constricted upper corolla tube in bud. In the 5S-NTS and the combined analyses, Alibertia garapatica occurs as their immediate sister. Apart from this clade the 5S-NTS and the combined analyses show several clades that are strongly supported by the bootstrap analyses. One comprises Alibertia amplexicaulis and A. aff. stricta, which is supported in 100% of the bootstrap replicates. Alibertia sessilis is their immediate sister taxon, supported in the 5S and the combined analyses 100% by the bootstrap. Among noteworthy characters supporting these three taxa is a 180-bp long deletion in the aligned 5S matrix. Alibertia elliptica, A. concolor, and A. hassleriana form another monophyletic group, which is supported in 93 and 81% of bootstrap replicates in the 5S and the combined analyses, respectively.
CONCLUSIONS
Although the tree topologies of the ITS and 5S analyses are largely congruent, the present study shows that the 5S nontranscribed spacer is a better tool than the ITS region for inferring relationships among closely related taxa in the Alibertia group. The informative characters are about twice as many, the resolution is greater, and the CI and RI values are higher. Analyses of both data sets separately and in conjunction have demonstrated several new hypotheses of relationship in the Alibertia group. Several groups are strongly supported in both the 5S-NTS and ITS analyses as well as in the combined analysis. However, all analyses indicate different weakly supported relationships of the basal branches, suggesting that the substitution rate was not appropriate for phylogeny estimation at this level.
The genus Alibertia can be subdivided in two well-defined and strongly supported lineages. One comprises several Alibertia species (including the type species, A. edulis), three representatives of the genus Borojoa, and Randia tessmannii, whereas the other group consists entirely of species that have traditionally been referred to Alibertia. The first group is mainly confined to humid rainforest areas and can be distinguished morphologically from the latter group by the possession of relatively larger fruits and corollas, higher corolla merosity and porate vs. colporate pollen grains.
The two Amaioua species appear in all analyses nested inside a strongly supported Duroia clade, suggesting that Duroia is paraphyletic as presently circumscribed. In two of the analyses, Amaioua is monophyletic and Duroia hirsuta comes out as the sister to rest in the Duroia clade, suggesting that Karsten's placement of D. hirsuta in a separate genus had some reason. However, A. guianensis appears on the same clade as Duroia hirsuta in the 5S-NTS analysis, and overall support for internal branches is weak.
The close relationship of Kutchubaea and Ibetralia, demonstrated by Persson's cpDNA analyses (C. Persson, in press) is confirmed in the present study. In addition, Alibertia hispida belongs to this alliance. Finally, the inclusion of Genipa aff. williamsii and the undescribed Colombian species in the Alibertia group data is supported in the present analyses.
FOOTNOTES
1 The author thanks Victor Albert, Lennart Andersson, and Roger Eriksson for constructive criticism of the manuscript and proofreading of the data matrices; Johan Rova for proofreading of the matrices as well as technical assistance; Laura Di Laurenzio, Lena Struwe, Eva Wallander, and Vivian Aldén for technical support; Piero Delprete for making herbarium material available at the New York Botanical Garden; Stephan G. Beck and Monica Moraes for their help during a field trip in Bolivia; and Claes Gustafsson, Bertil Ståhl, and Bian Tan (Strybing Arboretum, San Francisco) who provided plant material for sequencing. Financial support received from the Lewis B. and Dorothy Cullman Foundation, Adlerbertska forskningsfonden, Kungliga Hvitfeldtska stipendieinrättningen, Helge Ax:son Johnsons stiftelse, A. F. Regnell botaniska gåvomedel, and P. A. Larsson's stipendiefond is gratefully acknowledged. 
3 Author for correspondence (e-mail: claes.persson{at}systbot.gu.se
).
LITERATURE CITED
Andersson, L. 1992 A provisional checklist of neotropical Rubiaceae. Scripta Botanica Belgica 1: 1–200.
Andreasen, K., B. G. Baldwin, and B. Bremer. 1997 Phylogenetic utility of nuclear rDNA ITS region in the subfamily Ixoroideae (Rubiaceae): comparisons with cpDNA rbcL sequence data. In K. Andreasen, Phylogeny of Ixoroideae: molecular systematics within Rubiaceae. Ph.D. dissertation, Uppsala University, Uppsala, Sweden.
———, and B. Bremer. 1996 Phylogeny of the subfamily Ixoroideae (Rubiaceae). Opera Botanica Belgica 7: 119–138.
———, and ———. 1997 Combined phylogenetic analysis in the Rubiaceae-Ixoroideae: morphology and chloroplast DNA data. In K. Andreasen, Phylogeny of Ixoroideae: molecular systematics within Rubiaceae. Ph.D. dissertation, Uppsala University, Uppsala, Sweden.
Baldwin, B. G. 1993 Molecular phylogenetics of Calycadenia (Compositae) based on ITS sequences of nuclear ribosomal DNA: chromosomal and morphological evolution re-examined. American Journal of Botany 80: 222–238.[CrossRef][Web of Science]
———, M. J. Sanderson, J. M. Porter, M. F. Wojciechowskii, C. S. Campbell, and 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][Web of Science]
Bremekamp, C. E. B. 1934 Notes on the Rubiaceae of Surinam. Medeelingen van het Botanisch Museum en herbarium van de Rijksuniversiteit te Utrecht 11: 248–308.
Buckler IV, E. S., A. Ippolito, and T. P. Holtsford. 1997 The evolution of ribosomal DNA: divergent paralogues and phylogenetic implications. Genetics 145: 821–832.[Abstract]
Cox, A. T., M. D. Bennett, and T. A. Dyer. 1992 Use of polymerase chain reaction to detect spacer size heterogeneity in plant 5S-rRNA gene clusters and to locate such clusters in wheat (Triticum aestivum L.) Theoretical and Applied Genetics 83: 684–690.[Web of Science]
Cronn, R. C., X. Zhao, A. H. Paterson, and J. F. Wendel. 1996 Polymorphism and concerted evolution in a tandemly repeated gene family: 5S Ribosomal DNA in diploid and alloploid cottons. Journal of Molecular Evolution 42: 685–705.[CrossRef][Web of Science][Medline]
Cuatrecasas, J. 1948 Borojoa, un nuevo género de Rubiaceas. Revista de la Academia Colombiana de Ciencias Exactas, Físicas-Químicas y Naturales 7: 474–481.
———. 1953 Caracterización del género Borojoa. Acta Agronómica 3:89–98
Delprete, P. G. 1999 Riodocea (Rubiaceae, Gardenieae) a new genus from the Brazilian Atlantic forest. Brittonia 51: 15–23.[CrossRef][Web of Science]
Donoghue, M. J., and M. J. Sanderson. 1992 The suitability of molecular and morphological evidence in reconstructing plant phylogeny. In P. S. Soltis, D. E. Soltis and J. J. Doyle [eds.], Molecular systematics of plants, 340–368. Chapman and Hall, New York, New York, USA.
Doyle, J. J. 1992 Gene trees and species trees: molecular systematics as one-character taxonomy. Systematic Botany 17: 144–163.[CrossRef][Web of Science]
———, and J. L. Doyle. 1987 A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochemical Bulletin 19: 11–15.
Hillis, D. M., and M. T. Dixon. 1991 Ribosomal DNA: Molecular evolution and phylogenetic inference. Quarterly Review of Biology 66: 411–453.[CrossRef][Medline]
Huelsenbeck, J. P., J. J. Bull, and W. Cunningham. 1996 Combining data in phylogenetic analysis. Trends in Ecology and Evolution 11: 152–158.[CrossRef]
Karsten, H. 1859 Plantae Karstenianae. Linnaea—ein Journal für die Botanik in ihrem ganzen Umfange 30: 143–184.
Kellogg, E. A., and R. Appels. 1995 Intraspecific and interspecific variation in 5S RNA genes are decoupled in wheat relatives. Genetics 140: 325–343.[Abstract]
Lorence, D. H. 1986 Glossostipula (Rubiaceae) a new genus from Mexico and Guatemala. Candollea 41: 453–461.
Nickrent, D. L., K. P. Schuette, and E. M. Starr. 1994 A molecular phylogeny of Arceuthobium (Viscaceae) based on nuclear ribosomal DNA internal transcribed spacer sequences. American Journal of Botany 81: 1149–1160.[CrossRef][Web of Science]
Olmstead, R. G., and J. A. Sweere. 1994 Combining data in phylogenetic systematics: an empirical approach using three different data sets in the Solanaceae. Systematic Biology 43: 467–481.[CrossRef][Web of Science]
Oxelman, B., M. Lidén, and D. Berglund. 1997 Chloroplast rps16 intron phylogeny of the tribe Sileneae (Caryophyllaceae). Plant Systematic and Evolution 206: 393–410.[CrossRef]
Persson, C. 1993 Pollen morphology of the Gardenieae-Gardeniinae (Rubiaceae). Nordic Journal of Botany 13: 561–582.[Web of Science]
———. 1995 Exotesta morphology in Gardenieae-Gardeniinae (Rubiaceae). Nordic Journal of Botany 15: 285–300.[Web of Science]
———. 1996 Phylogeny of Gardenieae (Rubiaceae). Botanical Journal of Linnean Society 121: 91–109.[CrossRef]
———. In press. Phylogeny of Gardenieae (Rubiaceae) based on chloroplast DNA sequences from the rps16 intron and trnL(UAA)-F(GAA) intergenic spacer. Nordic Journal of Botany.
Playford, J., R. Appels, and B. R. Baum. 1992 The 5S units of Acacia species (Mimosaceae). Plant Systematic and Evolution 183: 235–247.[CrossRef]
Robbrecht, E., and C. Puff. 1986 A survey of the Gardenieae and related tribes (Rubiaceae). Botanische Jahrbücher für Systematik, Pflanzengeschichte und Pflanzengeographie 108: 63–137.
Sastri, D. C., K. Hilu, R. Appels, E. S. Lagudah, J. Playford, and B. R. Baum. 1992 An overview of the evolution in plant 5S DNA. Plant Systematic and Evolution 183: 169–181.[CrossRef]
Schultze-Motel, J. 1986 Rubiaceae. In R. Mansfeld [ed.], Verzeichnis landwirtschaftlicher und gärtnerischer Kulturpflanzen (ohne Zierpflanzen) Band 2. Springer-Verlag, Berlin, Germany.
Schumann, K. 1891 Rubiaceae. In A. Engler and K. Prantl [eds.], Die natürlichen Pflanzenfamilien 4(4): 1–154. Engelmann, Leipzig, Germany.
Struwe, L., M. Thiv, J. W. Kadereit, A. S.-R. Pepper, T. J. Motley, P. J. White, J. H. E. Rova, K. Potgieter, and V. A. albert. 1998 Saccifolium (Saccifoliaceae), an endemic of sierra de la Neblina on the Brazilian-Venezuelan Border, is related to a temperate-alpine lineage of Gentianaceae. Harvard Papers of Botany 3: 199–214.
Swofford, D. L. 1991 When are phylogeny estimates from molecular data incongruent? In M. M. Miyamoto and J. Cracraft [eds.], Phylogenetic analysis using DNA sequences, 295–333. Oxford University Press, New York, New York, USA.
———. 1993 Phylogenetic Analysis Using Parsimony, version 3.1.1 Computer program distributed by Illinois Natural History Survey, Champaign, Illinois, USA.
White, T. J., T. Bruns, S. Lee, and J. Taylor. 1990 Amplification and direct sequencing of fungal ribosomal RNA genes for phylogenetics. In M. A. Innis, D. H. Gelfund, J. J. Sninsky, and T. J. White [eds.], PCR protocols: a guide to methods and applications, 315–322. Academic Press, San Diego, California, USA.
This article has been cited by other articles:
|
|
 |
|
  D. R. Morgan, R.-L. Korn, and S. L. Mugleston Insights into reticulate evolution in Machaerantherinae (Asteraceae: Astereae): 5S ribosomal RNA spacer variation, estimating support for incongruence, and constructing reticulate phylogenies Am. J. Botany, May 1, 2009; 96(5): 920 - 932. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. H. Mitchell and P. K. Diggle The evolution of unisexual flowers: morphological and functional convergence results from diverse developmental transitions Am. J. Botany, July 1, 2005; 92(7): 1068 - 1076. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. D. Alejandro, S. G. Razafimandimbison, and S. Liede-Schumann Polyphyly of Mussaenda inferred from ITS and trnT-F data and its implication for generic limits in Mussaendeae (Rubiaceae) Am. J. Botany, March 1, 2005; 92(3): 544 - 557. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. G. Razafimandimbison and B. Bremer Phylogeny and classification of Naucleeae s.l. (Rubiaceae) inferred from molecular (ITS, rBCL, and tRNT-F) and morphological data Am. J. Botany, July 1, 2002; 89(7): 1027 - 1041. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. H. E. Rova, P. G. Delprete, L. Andersson, and V. A. Albert A trnL-F cpDNA sequence study of the Condamineeae-Rondeletieae-Sipaneeae complex with implications on the phylogeny of the Rubiaceae Am. J. Botany, January 1, 2002; 89(1): 145 - 159. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
