| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Systematics and Phytogeography |
2Institute for Systematic Botany, University of Zurich, Zollikerstrasse 107, CH-8008 Zurich, Switzerland; 3Department of Botany, University of Cape Town, Rondebosch 7700, South Africa
Received for publication December 5, 2005. Accepted for publication April 20, 2006.
ABSTRACT
The evolutionary origins of the morphological and taxonomic diversity of angiosperms is poorly known. We used the genus Melianthus to explore the diversification of the southern African flora. Melianthus comprises eight species, and a phylogeny based on one nuclear and two plastid genes, as well as a morphological data set, confirmed that the genus is monophyletic. The two earliest diverging lineages are found in relatively mesic habitats, whereas the two terminal clades (an eastern and a western clade), each with three species, favor more arid habitats. The eastern clade is largely restricted to the summer-rainfall parts of southern Africa, and the western clade is found in winter-rainfall region. Molecular dating indicates a mid-Tertiary origin of the genus, with diversification of the eastern and western clades coincident with the Late Miocene–Pliocene uplift of the Escarpment mountains and the establishment of summer aridity along the west coast. The remarkably complex flowers are indicative of sunbird pollination, but many smaller birds can also visit. Speciation may be the consequence of allopatric divergence into edaphic–climatic niches. Divergence in flower and inflorescence morphology might be in response to the divergent pressures for nectar conservation in arid regions coupled with the need for signaling to avian pollinators in generally shrubby vegetation.
Key Words: bird pollination • Cape flora • diversification • Melianthaceae • Melianthus • molecular dating • South Africa • speciation
Southern Africa has a rich flora of 20 955 seed plant species (Germishuizen and Meyer, 2003
), making up some 8% of the world's flora and over 40% of the African flora (Linder et al., 2005
). The region embraces a diversity of environments varying dramatically in terms of topography, altitude, rainfall volume and seasonality, and geology. Overlain on this environmental diversity is a series of distinctive floras, which include the Mediterranean-type flora of the Cape, the Afromontane flora of the eastern Escarpment, the succulent desert flora of the arid west, the steppe grassland of the central plateau, the coastal forest flora of the east coast, and the savanna flora to the north (van Wyk and Smith, 2001
).
The origins of the modern day floristic diversity of the region remain poorly understood. Various dramatic vegetation changes have been proposed for the Cenozoic (Axelrod and Raven, 1978
), but the underlying evidence is poor. Although it is apparent that the region experienced a series of climatic oscillations during the Quaternary (Scott, 1999
; Scott and Vogel, 2000
), it is not clear how these affected floristic evolution. One approach to understanding floristic change is through the historical study of the individual taxa making up the flora of interest. For example, there has been extensive research on the evolutionary history of several clades belonging to the Cape flora (Manning and Linder, 1992
; Linder and Mann, 1998
; Goldblatt et al., 2002
; Verboom, Linder, and Stock, 2003
; Verboom et al., 2004
; Bakker et al., 2005
), and the results have been used to develop a scenario of a floristic radiation into a newly emergent winter-rainfall region during the Neogene (Goldblatt and Manning, 2002
; Linder and Hardy, 2004
; Linder, 2005
). In contrast to the extensive work done on the Cape, there have been few attempts to reconstruct the evolution of lineages with broader distributions in southern Africa. Archibald et al. (2005)
showed that Zaluzianskya arose in the arid parts of southern Africa and subsequently adapted to more mesic conditions in the east, and Mummenhoff et al. (2005)
suggested that the seasonal distribution of rainfall influenced differentiation in Heliophila.
In this study, we investigate the evolution of Melianthus, a genus of eight species belonging to the Melianthaceae. Melianthaceae includes five morphologically quite divergent genera: three in Africa and two in South America. Melianthus has a broad distribution in southern Africa, occupying a wide range of habitats and is represented in several centres of floristic endemism (e.g., Gariep, Kamiesberg, Hantam-Roggeveld, Albany, Drakensberg Alpine, Cape Floristic Region [van Wyk and Smith, 2001
]). Because six of the eight species have narrow distribution ranges, the genus has the potential to yield valuable insights into the evolution of the southern African flora. Unlike its sister genus, Bersama, which is entirely arborescent, Melianthus is composed entirely of shrubs (Fig. 1). The genus has several unusual features, including the production of black nectar by some species and highly foetid foliage. The latter has led to widespread ethnobotanical use (Watt and Breyer-Brandwijk, 1962
; Kelmanson et al., 2000
). There has been extensive phytochemical research on the genus (Anderson, 1968
; Anderson and Koekemoer, 1968
, 1969
; Koekemoer et al., 1970
, 1971
, 1974
; Agarwal and Rastogi, 1976
). Several species are also horticulturally interesting (Andrews, 1987
).
|
MATERIALS AND METHODS
DNA sequence data
Twenty accessions representing all species and subspecies of Melianthus and four outgroups (Greyia flanaganii, G. radlkoferi, Bersama lucens, B. swinnyi) were sampled from plants growing in the field or in cultivation (Appendix). Bersama is sister to Melianthus (Savolainen et al., 2000
), while Greyia, though being more distantly related to Melianthus, is nonetheless included with the South American genera Francoa and Tetilla in Melianthaceae (APG, 2003
). Voucher specimens of all accessions are housed in the Bolus Herbarium, University of Cape Town (BOL).
Total genomic DNAs were isolated from fresh leaves or from leaves collected into a 2xCTAB solution containing 1% polyvinylpyrrolidone (PVP). Extractions were done according to the protocol of Gawel and Jarrett (1991)
, with 0.05–0.1 mg PVP powder added during grinding.
Three noncoding DNA loci representing both plastid and nuclear genomes were sampled using PCR. The trnL-trnF region (including both the trnL intron and the trnL-trnF intergenic spacer) and the psbA-trnH intergenic spacer of the plastid genome were amplified, respectively, using primers c and f (Taberlet et al., 1991
), and primers psbAF and trnHR (Sang et al., 1995
). The nuclear internal transcribed spacer (ITS) region was amplified using primers ITS5 and ITS4 (White et al., 1990
). Amplification reactions (50 µL) contained 5 mM MgCl2, 1x DNA polymerase reaction buffer (Bioline, London, UK), 0.4 mM total dNTP, 0.2 µM each primer, 0.05 units/µL Taq DNA polymerase (Bioline), and 2 µL of template DNA. For ITS, reactions also contained 2% dimethyl sulfoxide (Buckler et al., 1997
). Amplifications were done on a GeneAmp PCR System 2700 (Applied Biosystems, Foster City, California, USA) with the following thermal profile: 2 min at 94°C; 30 cycles of 1 min at 94°C, 1 min at 52°C (plastid loci) or 55°C (ITS), and 2 min at 72°C; and 7 min at 72°C. PCR products were cleaned using a Qiaquick PCR Purification Kit (Qiagen GmBH, Hilden, Germany) and cycle-sequenced using the ABI PRISM BigDye Terminator Cycle Sequence Kit (Applied Biosystems, Warrington, UK). Sequence reactions were done using the amplification primers and, in some cases, an additional internal primer (for trnL-trnF primer e, and for ITS primers ITS3 (White et al., 1990
) or 5.8S (Hershkovitz and Lewis, 1996
)) was used. Cycle sequencing reactions (10 µL) contained 2.0 µL of BigDye (version 3.0), 0.5x cycle sequence buffer (Applied Biosystems, Warrington, UK), 0.16 µM primer, and 0.5–2.0 µl of PCR product. For ITS, reactions also contained 1% dimethyl sulfoxide. Reactions were run over 25 cycles, each comprising 30 s at 96°C, 15 s at 50°C, and 4 min at 60°C, and the products were run out on an ABI PRISM 3100 Genetic Analyser (Applied Biosystems, Foster City, California, USA).
Forward and reverse sequences were assembled and edited in SeqMan II version 2.04 (Lasergene Software, DNASTAR, Madison, Wisconsin, USA) before being manually aligned in MegAlign (Lasergene Software). Although alignment required the insertion of several gaps, assessment of sequence homology was largely unambiguous. Four hypervariable regions in the ITS sequences of Greyia (corresponding to positions 63–80, 94–127, 169–231, and 409–442 in the ITS sequence of G. flanaganii) were exceptional in this regard, proving impossible to align meaningfully to those of Melianthaceae. Rather than discard these sites (which were alignable within Melianthaceae), we coded them as unknown for the Greyia outgroups.
For phylogenetic analysis, nucleotide gaps were treated as missing, and the presence/absence of insertions/deletions (indels) coded separately using simple gap coding (Simmons and Ochoterena, 2000
) as implemented in the program GapCoder (Young and Healy, 2003
). Although most indels so scored were included in subsequent analyses, those flanking long single-base repeats (corresponding to positions 429–430 in the ITS of M. comosus 4; positions 184–203 in the psbA-trnH of M. comosus 4; and positions 657–666 in the trnL-trnF of M. comosus 4) were excluded due to scoring ambiguity. A 30–31 bp stretch in the psbA-trnH sequences (corresponding to positions 102–132 in the psbA-trnH sequence of M. comosus 4) containing an inversion in some taxa was also excluded from analyses, the presence/absence of the inversion being scored as a single binary character.
The final molecular character matrix, excluding ambiguous gaps and the inversion sequence, comprised a total of 2142 characters (299 informative) consisting of 1430 plastid (133 informative) and 712 (166 informative) nuclear DNA characters.
Morphological data
Morphological characters were scored from 243 Melianthus specimens drawn from the following herbaria: B, BOL, E, GRA, K, NU, PRE, SRGH, and WIND (abbreviations follow Holmgren and Holmgren, 1998
onwards). In addition, most taxa were also investigated in the field, in many cases at several populations.
Seed morphology was investigated using field-collected seed as well as seed acquired from Silverhill Seed Company or the seed bank at the National Botanic Gardens, Kirstenbosch, South Africa. Ten seed length and width measurements per collection were measured using a dissecting microscope fitted with an eyepiece graticule. To study seed coat micromorphology, dry seeds were cut into pieces c. 2 x 2 mm, mounted directly on stubs without critical-point drying, then sputter coated with gold and and photographed using a Cambridge S200 scanning electron microscope (SEM) at 10 kV.
Palynological variation was studied using pollen grains acetolyzed according to the method of Erdtman (1952)
, modified as follows: anthers were excised, soaked overnight in a wetting agent, washed in glacial acetic acid, and digested in an acetolysis mixture (9:1 glacial acetic acid to sulphuric acid). Digestion was achieved by gently heating the mixture in a boiling water bath in a fume cupboard for 10 min. Acetic acid provides a suitable interface between water and acid and stains the pollen grains; consequently, it was used to wash the grains before they were rinsed in distilled water. For light microscopy, acetolyzed pollen grains were mounted in glycerine jelly and fixed semipermanently in paraffin wax. They were observed using a Zeiss Standard 25 light microscope and photographed on a Zeiss Axioskop microscope using bright field optics. For SEM, acetolyzed grains were mounted on stubs, dried by allowing the 70% alcohol they were suspended in to evaporate, and sputter coated with gold, before being photographed.
The final morphological character matrix comprised 54 characters, all parsimony informative (Appendices S1 and S2, see Supplemental Data accompanying the online version of this article). The data matrix and resulting trees have been submitted to TreeBase (SN 2741) at http://www.treebase.org/treebase/.
Phylogenetic analyses
The plastid (trnL-trnF + psbA-trnH) DNA, nuclear (ITS) DNA and morphological partitions were analyzed using parsimony, both separately and in combination. Two combined analyses were done, one using only the two molecular partitions and a second based on all three data partitions. For the latter analysis, the molecular data sets were pared down to include a single representative of each Melianthus species, as well as a single species each of Bersama and Greyia. All searches were conducted using the Branch and Bound search algorithm implemented in PAUP* version 4.0b10 (Swofford, 2002
). Characters were treated as unordered and equally weighted (Fitch, 1971
). Character support for nodes was evaluated using 1000 bootstrap replicates (Felsenstein, 1985
). Bootstrap searches were conducted using a heuristic search procedure with a simple addition sequence and with MAXTREES set to 1000.
Molecular dating
We tested whether the molecular evolution in Melianthus was clocklike by evaluating the likelihood of the total evidence topology, with and without a clock assumption, in the context of the combined three-gene data and a model determined as appropriate using Modeltest 3.6 (Posada and Crandall, 1998
). The resulting likelihoods were compared using a log-likelihood ratio test.
In the absence of a clock, the ages of speciation events were estimated using a relaxed Bayesian clock as implemented in Multidivtime (Thorne and Kishino, 2002
) and following the protocols suggested by Rutschmann (2004)
. We selected the F84 model of molecular change. This model was parameterized by the baseml module of the program PAML (Yang, 1997
). The branch lengths for the chosen cladogram plus our data matrix, as well as a variance–covariance matrix, were estimated using Estbranches (distributed with Multidivtime). The tree was calibrated using the ages calculated by Wikström et al. (2001)
for the node linking Bersama with Greyia. Wikström et al. obtained a maximum age of 67 million years (My) for this node using ACCTRAN, and a minimum age of 59 My using ML, with a standard deviation of 7 My. We set these dates as upper and lower bounds, but did not take the standard deviations into account. These dates have an unknown error in them, and should be regarded as exceedingly rough estimates. Using as additional priors the number of time units between the tip and the root of the tree (rttm), the distribution of the rate at the root node, and a constant for the Brownian motion, we used Multidivtime to estimate the posterior distribution of the node ages. We ran the Markov Chain for 106 generations, with every 10th generation being sampled, and the first 10 000 samples being discarded. Rttm, the a priori expected number of time units between tip and root, was set to 1.5, and the mean of the prior distribution of the rate at the root node was set to 0.09. The latter was calculated by Estbranches as the mean amount of evolution from the root node to the tips, divided by the arbitrarily set rttm. Brownmean, the prior for the brownian motion constant, was set to equal the product of Brownmean and rttm, and to lie between 1 and 2 as suggested by Rutschmann (2004)
. Due to the relatively small size of the data set, the data were not partitioned but analyzed under a single model. The analysis was run twice and the results compared to establish whether convergence had occurred.
The hypothesis that there might have been a slowdown in the speciation rate was tested with the program Gammastatistic (Griebeler, 2004
).
Ecological reconstructions
Species distributions were mapped using herbarium specimen locality data, as well as field records made over several field seasons. Using these distributions, annual rainfall ranges for each species were abstracted from the rainfall models of the South African Weather Bureau (Schulze, 1997
), the ranges being coded categorically in steps of 200 mm, from 0 to 1200 mm. Rainfall seasonality was coded simply as "summer-dry" and "winter-dry." The bedrock associations of each species were coded according to the major bedrock types (e.g., granite, recent sand, shale, basalt, cave sandstone, or Table Mountain sandstone) according to the geological maps of southern Africa. Bedrock scoring was based on field observation, herbarium specimen label data, and inferences drawn on the basis of species' distribution ranges. To accommodate ancestral polymorphism, the data were coded categorically as presence-only (Hardy and Linder, 2005
). Both DELTRAN and ACCTRAN optimization, as implemented in MacClade 4.0, were used. Although this approach does not provide an estimate of the level of confidence in the resulting optimizations, it does cover the full range of equally parsimonious solutions.
Pollinator observations
Pollinator specificity in Melianthus was evaluated on the basis of observations made in the field during August–December 2002. During this period, all species except M. insignis and M. villosus were seen flowering in the field, and the full set of bird species visiting each investigated population of Melianthus was noted. Based on our observations, it was not possible to distinguish among bird species that are effective pollinators and those that are not. Using accounts in Maclean (1993)
, bird visitors were subsequently categorized as being (1) primarily nectarivorous, (2) occasionally nectarivorous, or (3) non-nectarivorous.
The nectar of all species except M. insignis was sampled, the sample for each species typically being drawn from 10 individuals (one flower per individual) belonging to a single population. The only exception was M. villosus for which the 10 replicates came from a single multibranched specimen in cultivation. Nectar was extracted with a 10–100 µL Finnpipette micropipette (Labsystems, Helsinki, Finland), this also being used to estimate volumes. Sucrose readings were calculated using a hand-held refractometer (Atago, 32–10 Honcho, Itubashi-ku, Japan), which measures nectar concentration in terms of equivalent percentage sucrose ratios.
RESULTS
Phylogeny
Consensus topologies obtained from the various parsimony analyses are shown in Fig. 2 with associated tree statistics listed in Table 1. The levels of homoplasy are remarkably low.
|
|
Within Melianthus, identical relationships are resolved by the combined molecular and total evidence data sets (Fig. 2d). Largely on the strength of the plastid data, M. major (Fig. 1b, g) is identified as sister to the rest of Melianthus, within which M. villosus is resolved as sister to a clade comprising two clades, one comprising three species from the western part of southern Africa (western clade) and the other comprising three species centered on the eastern and central plateau, although M. comosus reaches the coastal mountains both in the southern Cape and along the western escarpment (eastern clade). Support levels are generally strong, except for the grouping of the western and eastern clades.
Molecular dating
The molecular clock was rejected. Replicate dating runs using a relaxed Bayesian clock produced very similar results, suggesting that both runs reached stationarity. The result of one of these analyses is shown in Fig. 3 with the results summarized in Table 2. Partly due to our reliance on a single calibration point, the variances surrounding our age estimates are large. In general, the data indicate an Oligocene divergence between Melianthus and Bersama, followed by an early Miocene differentiation of M. major and M. villosus. The eastern and western clades diverged around 15 Mya. Within these two clades, the bulk of speciation happened prior to 8 Mya, though there is no significant slow-down in the diversification rate.
|
|
|
Pollination and nectar volumes
Field observations suggests that most species of Melianthus are visited by a broad range of bird species, and many avian visitors, especially nectarivores and occasional nectarivores, visit several Melianthus species (Table 3). This is certainly true for the most ubiquitous visitors to Melianthus, such as the lesser double-collared sunbird Cinnyris chalybeus and Cape whiteye Zosterops pallidus. For example, at a site near Montagu in the western Cape, where M. major and M. comosus co-occur, observations made over a period of 33 h suggest that the Z. pallidus accounts for about 50% of visits to flowers of both species.
|
Nectar volumes varied considerably among the species (Table 4), being greatest in M. major (81.0 µL), M. villosus, and M. dregeanus and lowest in M. gariepinus (14.6 µL). Sucrose concentration varied less than nectar volume, ranging between 9.7% and 13.5%.
|
Phylogeny
Our data strongly support the monophyly of Melianthus and provide a well-resolved and strongly supported phylogenetic hypothesis of species relationships within the genus. Although there is some conflict among the data partitions on which our analyses are based, this conflict is consistently weak, suggesting that it is the combined result of limited character sampling and stochastic sequence variation (Mason-Gamer and Kellogg, 1997; Eldenäs and Linder, 2000).
The magnitude of support for the monophyly of Melianthus is unsurprising: the genus is very different from its relatives, Bersama and Greyia, and has previously even been separated at family level (Doweld, 2001
). The latter two genera are tropical trees, with woody trunks. Although Greyia has, like Melianthus, showy, red, bird-pollinated flowers, these are actinomorphic (Dahlgren and van Wyk, 1988
; Ronse Decraene and Smets, 1999
). Bersama has zygomorphic flowers, with nectaries that are, as in Melianthus, sometimes unilateral (Verdcourt, 1950
). However, the flowers are relatively small. Within Melianthus, the identification of M. major and M. villosus as early-diverging elements suggests that the large, long-stalked, terminal inflorescences (Fig. 1b, c), that are held high above the large leaves, represent the plesiomorphic form in the genus. Also, the presence of highly distinctive, single stipules in M. major as well as Bersama suggests that these, rather than the more common paired stipules represent the primitive state. Solitary stipules are unusual in angiosperms as a whole (Roth, 1949
).
The remainder of the genus is divided into two clades. The western clade, with three species restricted to the western margin of southern Africa, is distinguished by having the inflorescences capped by a colorful crown of sterile flowers (Fig. 1a, h, i), and by acutely lobed fruit. In this lineage, as in M. major and M. villosus, the inflorescences are erect and borne above the foliage; however, their stalks are not as long stalked and they are therefore less prominent. In contrast, the eastern clade, with three species from the central and eastern part of southern Africa, is diagnosed by having "nodding," axillary inflorescences, that are borne beneath the apical leaves (Fig. 1c). Furthermore, the sepals have large distinctive red to black blotches near the base (Fig. 1f).
Timescale of diversification
Molecular dating is subject to error from a diversity of sources (Bell and Donoghue, 2005
), some of which may affect our efforts to date divergence times in Melianthus. First, accurate dating requires that the underlying phylogenetic hypothesis is correct and that this phylogeny is tracked by all the genes sampled. This is probably true for our study because our phylogeny is robust and the limited incongruence amongst our DNA data partitions is weakly supported and most likely reflects stochasticity in sequence variation. Second, under rate correction, nodal age estimation may be influenced by incomplete species sampling (Linder et al., 2005
). This may be problematic given our scant sampling of Bersama and Greyia. However, this problem may be largely offset by our use of a relaxed Bayesian clock, which has been found to introduce only minor distortions compared with some other methods (Linder et al., 2005
). Third, use of an under parameterized model of sequence evolution could systematically distort branch lengths from the tips to the root node (Sanderson, 1990
). Although we do not offer a comprehensive solution to this problem, we have minimized the risk of such distortion by selecting the F84 model, the most complex implemented in Multidivtime. The final source of error is calibration error (Conti et al., 2004
; Heads, 2005
). In our study, this error is a potential problem because we lack fossil references and so rely exclusively on secondary calibration points. The use of secondary calibrations can result in compounded error, with the result that variances on age estimates are greatly increased (Graur and Martin, 2004
; Hedges and Kumar, 2004
). The ages estimated by Wikström et al. (2001)
are derived from a single calibration point and may contain substantial error, the magnitude of which is unknown. Although we have used both their minimum and maximum age estimates for our calibration node, we have not take into account any error on these ages. Consequently, the substantial variances on our age estimates are almost certainly narrower than they should be (Table 2).
Notwithstanding the limitations involved, our data provide a broad, if somewhat tentative, timescale for the diversification of Melianthus (Fig. 3). The differentiation of Melianthus from Bersama dates from the Oligocene, when climates may have been similar to the modern, with extensive glaciation of Antarctica (Zachos et al., 2001
). Within Melianthus, the differentiation of M. major and M. villosus followed in the early Miocene, a wetter and warmer period than the Oligocene, and one in which the eastern Escarpment was uplifted by about 250 m (Partridge and Maud, 2000
). The emergence of the western and eastern clades is dated to around 15 My, almost contemporaneous with the initiation of further glaciation on Antarctica (Zachos et al., 2001
), which subsequently led to a drop in sea surface temperatures, strengthening of the south Atlantic high pressure cell, and consequent establishment of a distinct winter rainfall regime along the southern African west coast (Linder, 2003
). A continued drop in sea surface temperatures (Zachos et al., 2001
) through the late Miocene most likely accentuated the east–west climatic gradient. Within Melianthus, most speciation was completed by 8 My, though M. dregeanus and M. insignis diverged later, possibly in association with a second, more substantial phase of uplift (700–900 m) during the Pliocene (Partridge and Maud, 2000
).
Ecological background
The appearance of Melianthus may be tied to the cool, dry conditions of the Oligocene. Although there are few records of the climate and continental vegetation of this period, and none from southern Africa (Scott et al., 1997
), there are some indications of seasonal aridity (Leopold et al., 1992
). These conditions may have provided a selective advantage to perennial herbs occupying areas with groundwater, like stream-margins and damp hollows. In this context, the ability to die back to a persistent rootstock, a trait found in both M. major and M. villosus, would be advantageous, allowing plants to resprout when conditions were favorable. This may have been a more robust strategy than the plesiomorphic treelike habit found in Bersama and Greyia.
Our reconstructions of ancestral rainfall habitats imply contrasting patterns of arid adaptation in Melianthus. ACCTRAN suggests that the genus adapted to aridity early in its history, with later switches to moist environments, whereas DELTRAN projects a wetter habitat for the ancestral lineage, with adaptation to aridity occurring twice subsequently. However, the observed rainfall range for M. major probably misrepresents the water requirement of this species. Whereas M. major may be found in areas of quite low rainfall, it is restricted to streamlines and ditches where it has access to ground water. Consequently, it appears to require more moisture than is indicated by its rainfall score. If the rainfall range of M. major is rescored to reflect its high moisture requirement, then ACCTRAN and DELTRAN produce similar, high rainfall optimizations for the basal nodes. Thus, we infer that the genus arose in a high rainfall (or at least a damp) environment (600-1200 mm) and that the switch to aridity occurred three times: in M. major, in M. comosus, and in the ancestor of the western clade. Whereas the latter two lineages may be considered arid-adapted, the first is able to persist in arid conditions only through its exploitation of ground water.
The onset of warmer, moister conditions in the early Miocene, paired with uplift along the eastern Escarpment, appears to have provided the opportunity for M. villosus to evolve. The appearance of a new montane habitat, trapping moisture off the Indian Ocean allowed this species to escape the ground water niche occupied, for example, by M. major. Today, M. villosus is restricted to the mid-altitude belt of the Kwazulu-Natal-Lesotho Drakensberg, where it receives the highest rainfall volumes of any Melianthus species.
Rainfall seasonality varies markedly across southern Africa. Most of the region is characterized by wet summers and dry winters, but the western seaboard and the south coast receive their rain in winter (Schulze, 1997
). The transition between these rainfall zones is sharp and underlies a correspondingly pronounced floristic boundary (Linder, 2003
). Our reconstructions indicate that Melianthus arose in the summer rainfall zone and has persisted there, with four modern species being restricted to summer rainfall environments: M. villosus and the three eastern clade species. Within the eastern clade, differentiation of the arid-adapted M. comosus was probably powered by continental uplift during the Miocene. This would have blocked the passage of orographic rain to the central interior, generating a dry, summer rainfall environment. This is currently exploited by the widespread M. comosus, whereas the related M. dregeanus and M. insignis occupy moister environments near the Escarpment edge.
The divergence of the western clade from the eastern clade marks the occupation of summer-arid environments by Melianthus. The occupation of such environments is clearly derived in the genus and coincides temporally (ca. 15 My) with events leading to the establishment of a winter rainfall regime along the southern African west coast. Such a nested pattern, in which species restricted to arid, summer-dry conditions are nested within a clade characterized by wetter summer- or all-year rainfall conditions has already been documented for a number of taxa (Linder and Mann, 1998
; Verboom et al., 2003
; Bakker et al., 2005
). Although various lineages, most notably the ruschioid Aizoaceae (Klak et al., 2004
) have radiated spectacularly in the winter rainfall zone, Melianthus has yielded just three species there.
The use of bedrock to characterize the substrate in which plants grow is questionable because it is often a poor indicator of soil-type. However, optimization of bedrock type indicates a preference for granite, despite this not being the most common bedrock in southern Africa. The only species restricted to other bedrock types are M. villosus, which occurs on basaltic soils, and M. major, which is found on Table Mountain sandstone and shale (Fig. 1g).
Pollination
Melianthus has a typical avian pollination syndrome (Proctor et al., 1996
): vivid, red floral colors, large flowers with a hardened corona featuring stiff filaments (Fig. 1d–f), and abundant nectar of low sucrose concentration (9.7–13.5%). As nectarivorous African birds are incapable of hovering, the robust inflorescence stalks (Fig. 1a–c) are also typical. Most likely, the black color of the nectar of M. comosus and M. elongatus contributes to polychromatism in the flowers, forming the tricolor "parrot coloration" of Faegri and van der Pijl (1979)
.
We suggest that the pollination system of Melianthus is of the generalist ornithophilous type. Although sunbirds (Nectariniidae) are generally cited as the principal pollinators of Melianthus (Scott-Elliot, 1890
; Von Marilaun, 1895
; Burtt Davey, 1932
; Vogel, 1954
), many other avian visitors have been reported (Marloth, 1908
, 1925
; Skead, 1967
; Maclean, 1993
). Our field observations confirm these reports in showing that most species are visited by a broad array of bird species, including high proportions of the nectarivore species with which they are sympatric. Because our observational data are of an anecdotal nature, we cannot distinguish which bird visitors are effective in pollination and which are not. The shallow, open form of the flowers (Fig. 1d–f) suggests, however, that most bird species should be able to access the nectar, while the protrusion of the sexual parts and the versatile pedicels (which enable the flowers to tip downward) suggests that pollen should be effectively transferred to and from the heads of most birds that enter the flowers. This is especially true for shorter-billed birds such as the Cape whiteye, Zosterops pallidus, which is a particularly frequent visitor to most species of Melianthus, regardless of floral form. The observation that most Melianthus species are visited by a majority of sympatric nectarivores suggests, then, that the flowers do not filter bird visitors and instead employ a generalist bird pollination syndrome. We note that there are some instances where potential bird pollinators were not observed to visit particular Melianthus species, despite an overlap in distribution ranges. However, these are readily attributable to the limited scale of our sampling: our bird visitor data are based on observations at a limited number of populations of each Melianthus species, over just a single flowering season.
A generalist pollination syndrome suggests a limited role for pollinator selection in driving the floral evolution and speciation (Waser, 1998
). Pollinator specialization has been demonstrated in many southern African families and genera (Johnson and Steiner, 2000
, 2003
), commonly resulting in tight linkages between pollinator shifts, evolution of floral morphology, and speciation (Johnson, 1996
). A lack of evidence for pollinator specialization in Melianthus, then, poses two questions: what drives the diversification in floral and inflorescence structure in the genus, and what drives speciation?
Floral evolution
Although Melianthus has remarkably complex flowers, evidence for a generalist avian pollination system suggests that the substantial variation in floral and inflorescence structure may not be related to the attraction of specialist pollinators. Alternative explanations include the protection of nectar against desiccation and a general need for plants to attract avian pollinators.
Our data suggest that interspecific variation in nectar volume is related to the aridity of the habitats in which the different species occur (Fig. 5). We recorded the largest nectar volumes in M. major, which grows in wet hollows and seepages with perennial or near-perennial groundwater, and the smallest volumes in M. gariepinus from the arid southern Namib Desert. Our estimates of nectar production are, however, based on measurements from single populations and so fail to account for interpopulational variation in nectar production. Additionally, because our nectar readings were taken from unbagged flowers, it is possible that the observed differences are due to differential levels of nectar removal by bird visitors rather than intrinsic variation in nectar production among species. If the interspecific variation in nectar volume were the result of erratic nectar removal by birds, however, we would expect the species-specific standard errors to be broader than they are, especially because our replicate measurements were generally taken from flowers on different plants.
|
). Although consistently cuplike, they vary from being deep, with a high rim, to quite shallow. The shape of the nectary cup is related to the volume of nectar produced, with M. major and M. villosus having the deepest nectary cups. If nectar production is limited by water availability, changes in nectary shape and size may reflect adaptation to increasingly arid conditions. The shift of the inflorescence from a terminal position to below the leaves (Fig. 1c), as seen in the eastern clade, may be related to the protection of nectar against desiccation. In the western clade, the evolution of an unusual crown of sterile flowers (Fig. 1a) may serve to enhance pollinator attraction. These plants often occur scattered in the shrubby vegetation and, at least to the human eye, the crown of red, sterile petals greatly enhances their visibility (Fig. 1h, i). This may be necessary given the low nectar volumes generally produced by members of this clade.
We suggest that flower and inflorescence evolution have been jointly shaped by the need to optimize long-distance visibility to pollinators (competition with the surrounding vegetation) and protect the nectar against desiccation. This implies that these adaptations fine-tune the avian pollination syndrome to suit the ecology of the region.
Speciation
The apparent lack of pollinator specialization in Melianthus implies that the mechanisms underlying the genetic isolation associated with speciation have to be sought elsewhere. The highly restricted ranges of most species argues for an important role for allopatry, with climatic and edaphic factors playing a key role in selecting for novel morphologies.
Melianthus major and M. villosus occur in the wettest habitats of any Melianthus species, the former being associated with ground water and the latter with the wet Drakensberg Escarpment. Despite ages of 19 and 17 My, respectively, there has been no diversification in these two lineages. At least in the Cape flora, species-poor lineages are largely restricted to the wettest habitats, and these lineages are often regarded as relicts of previously more widespread clades (Levyns, 1962
; Goldblatt, 1997
; Linder, 2003
). However, these are also the geographically most restricted habitats and the interpretation that the small number of species is the result of physical space limitation cannot be rejected.
In Melianthus two radiations are evident in drier environments, both along the escarpment mountains, in what are topographically complex areas. Increased speciation in mountainous areas has already been demonstrated for the Americas (von Hagen and Kadereit, 2003
). In the case of the eastern clade, at least, speciation seems to have been associated with uplift events during the Miocene and Pliocene, these generating a series of sharp rainfall gradients on the eastern escarpment. By contrast, the emergence of the western clade was associated with adaptation to a winter rainfall regime. The species comprising both lineages are allopatric and in both lineages show some evidence of edaphic specialization. In the western lineage, M. elongatus has specialized to coastal Quaternary sands (Fig. 1h), whereas in the eastern clade, M. comosus is largely found on alluvial soils derived from shales. Speciation thus appears to have involved a combination of allopatry, edaphic specialization, and shifts in total rainfall. Such combinations have been repeatedly documented for diverse families in the Cape flora (Goldblatt, 1982
; Kurzweil et al., 1991
; Linder and Hardy, 2005
), but appear not to have been documented for the general southern African flora.
The low diversification rate in Melianthus, compared to many other southern African clades of a similar age (Klak et al., 2004
; Linder and Hardy, 2004
), is curious. We suggest that this could be the result of the unspecialized bird pollination system. Although bird pollination is fairly common in the southern African flora (Rebelo, 1987
), it is generally a derived syndrome, found in a few species in mostly insect-pollinated lineages (e.g., Johnson, 1996
; e.g., Manning and Goldblatt, 2005
). There has as yet been no investigation of the consequences to the speciation rate of a change to bird pollination. It is striking, though, that where the pollinators are known, biotically pollinated lineages with high diversification rates tend to have a remarkably broad range of pollination syndromes (Johnson et al., 1998
; Goldblatt et al., 2002
; Manning and Goldblatt, 2005
, 2006
), suggesting that being "locked" into a single pollination system might lead to lower diversification rates. Furthermore, bird pollination might result in longer gene-flow distances, thus inhibiting speciation.
It remains unclear whether the changes in Melianthus were in response to changing conditions or whether it gradually evolved the ability to deal with existing conditions. Paleoecological reconstructions suggest the former: in the middle Miocene southern Africa lacked dry, and most certainly summer-dry, habitats. Then the inferred ecological evolution in Melianthus reflects the climatic evolution of the subcontinent. If this is the case, we would expect many other taxa to have similar patterns, possibly reflecting a general model for the diversification of the southern African flora.
APPENDIX
Voucher information and GenBank accession numbers for ITS, trnL-trnF, and psbA sequences from taxa used in this study. Voucher specimens are deposited in the Bolus Herbarium of the University of Cape Town
Taxon: GenBank accession nos. ITS, trnL-trnF, psbA; Source (Acc. no.)
Bersama lucens (Hochst.) Szyszyl.: DQ435401, DQ435381, DQ435421; Cult. Kirstenbosch Bot. Gard. (439), J. Henning 34. B. swinnyi Phillips, DQ435402, DQ435382, DQ435422; Cult. Kirstenbosch Bot. Gard. (441), J. Henning 35.
Greyia flanaganii Bolus, DQ435399, DQ435379, DQ435419; Cult. Kirstenbosch Bot. Gard. (444), J. Henning 32. G. radlkoferi Szyszyl., DQ435400, DQ435380, DQ435420; Cult. Kirstenbosch Bot. Gard. (203), J. Henning 31.
Melianthus comosus Vahl 1, DQ435407, DQ435387, DQ435427; Namibia, Rosh Pinah, J. Henning 3. M. comosus Vahl 2, DQ435408, DQ435388, DQ435428; South Africa, Nieuwoudtville, J. Henning 34. M. comosus Vahl 3, DQ435409, DQ435389, DQ435429; South Africa, Montagu, J. Henning 21. M. comosus Vahl 4, DQ435410, DQ435390, DQ435430; South Africa, Golden Gate, J. Henning 25. M. comosus Vahl 5, DQ435411, DQ435391, DQ435431; South Africa, Zastron, J. Henning 27. M. elongatusWijn., DQ435418, DQ435398, DQ435438; South Africa, Velddrif, J. Henning 2. M. insignis (Phill. & Hoff.) Tansley and Schelpe, DQ435413, DQ435393, DQ435433; South Africa, Golden Gate, J. Henning 24. M. dregeanus (Sond.) Tansley and Schelpe, DQ435412, DQ435392, DQ435432; South Africa, Cathcart, J. Henning 29. M. major L. 1, DQ435403, DQ435383, DQ435423; South Africa, Montagu, J. Henning 1. M. major L. 2, DQ435404, DQ435384, DQ435424; South Africa, Cape Peninsula, J. Henning 20. M. major L. 3, DQ435405, DQ435385, DQ435425; South Africa, Citrusdal, J. Henning 16. M. pectinatus (Harv.) Tansley and Schelpe 1, DQ435414, DQ435394, DQ435434; South Africa, Steinkopf, J. Henning 9. M. pectinatus (Harv.) Tansley and Schelpe 2, DQ435415, DQ435395, DQ435435; South Africa, Kamiesberg, J. Henning 13. M. gariepinus(Merxm. & Roess.) Tansley and Schelpe 1, DQ435416, DQ435396, DQ435436; Namibia, Rosh Pinah, J. Henning 6. M. gariepinus(Merxm. & Roess.) Tansley and Schelpe 2, DQ435417, DQ435397, DQ435437; Namibia, Rosh Pinah, J. Henning 4. M. villosusBolus, DQ435406, DQ435386, DQ435426; Cult. Kirstenbosch Bot. Gard. (857), J. Henning 33.
FOOTNOTES
1 The authors thank the directors of B, E, GRA, K, NU, PRE, SRGH, and WIND for loaning their material to the Bolus Herbarium and Cape Nature and the Northern Cape Department of Nature Conservation for permission to collect material in areas under their jurisdiction, and gratefully acknowledge funding from the University of Cape Town and the NRF of South Africa. 
4 Author for correspondence (peter.linder{at}systbot.unizh.ch
; phone: +41 44 634 8410; fax: +41 44 634 8403
101 5 Current address: Peak Timbers Ltd., Piggs Peak, Swaziland.
102 6 Current address: Biology Department, Lehman–CUNY, New York 10468 USA
LITERATURE CITED
Agarwal J. S. Rastogi R. P.. 1976. Queretaroic (30) caffeate and other constituents of Melianthus major. Phytochemistry 15: 430-431.[CrossRef][ISI]
Anderson L. A.. 1968. Occurrence of oleanolic acid acetate and ellagic acid in Melianthus comosus Vahl. Journal of the South African Chemical Institute 21: 91-92.[ISI]
Anderson L. A. Koekemoer J. M.. 1968. Chemistry of Melianthus comosus Vahl. 2. Isolation of hellebrigenin 3-acetate and other unidentified bufadienolides from rootbark of Melianthus comosus Vahl. Journal of the South African Chemical Institute 21: 155-159.[Medline]
Anderson L. A. Koekemoer J. M.. 1969. Chemistry of Melianthus comosus Vahl. 3. Constitution of melianthugenin and melianthusigenin, new bufadienolides from root-bark. Journal of the South African Chemical Institute 22: 191-197.[Medline]
Andrews S.. 1987. Melianthus villosus. Kew Magazine 4: 123-127.
APG.. 2003. An update of the Angiosperm Phylogeny Group classification for the orders and families of flowering plants: APG II. Botanical Journal of the Linnean Society 141: 399-436.[CrossRef][ISI]
Archibald J. K. Mort M. E. Wolfe A. D.. 2005. Phylogenetic relationships within Zaluzianskya (Scrophulariaceae s.s., tribe Manuleeae): classification based on DNA sequences from multiple genomes and implications for character evolution and biogeography. Systematic Botany 30: 201-215.
Axelrod D. I. Raven P. H.. 1978. Late Cretaceous and Tertiary vegetation history of Africa. In M. J. A. Werger [ed.], Biogeography and ecology of southern Africa 79-130 Junk, The Hague, Netherlands.
Bakker F. T. Culham A. Marais E. Gibby M.. 2005. Nested radiation in Cape Pelargonium. In F. T. Bakker, L. W. Chatrou, B. Gravendeel, P. B. Pelser [eds.], Plant species-level systematics: new perspectives on pattern & process 75-100 Koeltz, Königstein, Germany.
Bell C. D. Donoghue M. J.. 2005. Dating the Dipsacales: comparing models, genes, and evolutionary implications. American Journal of Botany 92: 284-296.
Buckler E. Ippolito A. Holtsford T.. 1997. The evolution of ribosomal DNA: paralogues and phylogenetic implications. Genetics 145: 821-832.[Abstract]
Burtt Davey J.. 1932. A manual of flowering plants and ferns of the Transvaal and Swaziland Longmans, London, UK.
Conti E. Rutschmann F. Eriksson T. Sytsma K. J. Baum D. A.. 2004. Calibration of molecular clocks and the biogeographic history of Crypteroniaceae: A reply to Moyle. Evolution 58: 1874-1876.[ISI]
Dahlgren R. M. van Wyk A. E.. 1988. Structures and relationships of families endemic to or centered in southern Africa. Monogaphs of Systematic Botany from the Missouri Botanical Garden 25: 1-94.
Doweld A. B.. 2001. The systematic relevance of fruit and seed structure in Bersama and Melianthus (Melianthaceae). Plant Systematics and Evolution 227: 75-103.[CrossRef][ISI]
Erdtman G.. 1952. Pollen morphology and plant taxonomy Almqvist & Wiksell, Stockholm, Sweden.
Faegri K. van der Pijl L.. 1979. The principles of pollination biology Pergamon Press, Oxford, UK.
Felsenstein J.. 1985. Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39: 783-791.[CrossRef][ISI]
Fitch W. M.. 1971. Toward defining the course of evolution: minimal change for a specific tree topology. Systematic Zoology 20: 406-416.
Gawel N. Jarrett R.. 1991. A modified CTAB DNA extraction procedure for Musa and Ipomoea. Plant Molecular Biology Reporter 9: 262-266.
Germishuizen G. Meyer N. L.. 2003. Plants of southern Africa: an annotated checklist National Botanical Institute, Pretoria, South Africa.
Goldblatt P.. 1982. Systematics of Freesia Klatt (Iridaceae). Journal of South African Botany 48: 39-91.
Goldblatt P.. 1997. Floristic diversity in the Cape Flora of South Africa. Biodiversity and Conservation 6: 359-377.[CrossRef][ISI]
Goldblatt P. Manning J. C.. 2002. Plant diversity of the Cape Region of southern Africa. Annals of the Missouri Botanical Garden 89: 281-302.[CrossRef][ISI]
Goldblatt P.. 2006. Radiation of pollination systems in the Iridaceae of sub-Saharan Africa. Annals of Botany 97: 317-344.
Goldblatt P. Savolainen V. Porteous O. Sostaric I. Powell M. Reeves G. Manning J. C. Barraclough T. G. Chase M. W.. 2002. Radiation in the Cape flora and the phylogeny of peacock irises Moraea (Iridaceae) based on four plastid DNA regions. Molecular Phylogenetics and Evolution 25: 341-360.[Medline]
Graur D. Martin W.. 2004. Reading the entrails of chickens: molecular timescales of evolution and the illusion of precision. Trends in Genetics 20: 80-86.[CrossRef][ISI][Medline]
Griebeler E. M.. 2004. Gammastatistic, version 1.0 Institut für Zoologie, Abteilung Ökologie, Johannes Gutenberg-Universität, Mainz, Germany.
Hardy C. R. Linder H. P.. 2005. Reconstructing ancestral habitats and ecologies: accounting for intraspecific variability and issues of timing in ecological diversification. Systematic Biology 54: 299-316.[CrossRef][ISI][Medline]
Heads M.. 2005. Dating nodes on molecular phylogenies: a critique of molecular biogeography. Cladistics 21: 62-78.[ISI]
Hedges S. B. Kumar S.. 2004. Precision of molecular time estimates. Trends in Genetics 20: 242-247.[CrossRef][ISI][Medline]
Hershkovitz M. Lewis L.. 1996. Deep level diagnostic value of the rDNA-ITS region: the case of an algal interloper. Molecular Biology and Evolution 13: 167-177.
Holmgren P. K. Holmgren N. H.. 1998. onwards Index Herbariorum. New York Botanical Garden Website http://sciweb.nybg.org/science2/IndexHerbariorum.asp [Accessed 18 April 2006].
Johnson S. D.. 1996. Bird pollination in South African species of Satyrium (Orchidaceae). Plant Systematics and Evolution 203: 91-98.[CrossRef][ISI]
Johnson S. D.. 1996. Pollination, adaptation and speciation models in the Cape flora of South Africa. Taxon 45: 59-66.[CrossRef][ISI]
Johnson S. D. Steiner K. E.. 2000. Generalization versus specialization in plant pollination systems. Trends in Evolution and Ecology 15: 140-143.[CrossRef]
Johnson S. D. Steiner K. E.. 2003. Specialized pollination systems in southern Africa. South African Journal of Science 99: 345-348.[ISI]
Johnson S. D. Linder H. P. Steiner K. E.. 1998. Phylogeny and radiation of pollination systems in Disa (Orchidaceae). American Journal of Botany 85: 402-411.[Abstract]
Kelmanson J. E. Jager A. K. van Staden J.. 2000. Zulu medicinal plants with antibacterial activity. Journal of Ethnopharmacology 69: 241-246.[CrossRef][ISI][Medline]
Klak C. Reeves G. Hedderson T. A.. 2004. Unmatched tempo of evolution in Southern African semi-desert ice plants. Nature 427: 63-65.[CrossRef]
Koekemoer J. M. Anderson L. A. Pachler K. G. R.. 1970. Chemistry of Melianthus comosus Vahl. Part IV. 6-ß-acetoxymelianthugenin, a new toxic principle from root-bark. Journal of the South African Chemical Institute 23: 146-153.[ISI]
Koekemoer J. M. Anderson L. A. Pachler K. G. R.. 1971. Chemistry of Melianthus comosus Vahl. 5. 2 Novel bufadienolides from root-bark. Journal of the South African Chemical Institute 24: 75-86.[ISI]
Koekemoer J. Vermeulen N. M. J. Anderson L. A.. 1974. Chemistry of Melianthus comosus Vahl. 6. Structure of a new triterpenoid acid from root-bark. Journal of the South African Chemical Institute 27: 131-136.[ISI]
Kurzweil H. Linder H. P. Chesselet P.. 1991. The phylogeny and evolution of the Pterygodium–Corycium complex (Coryciinae, Orchidaceae). Plant Systematics and Evolution 175: 161-223.[CrossRef][ISI]
Leopold E. B. Liu G. Clay-Poole S.. 1992. Low-biomass vegetation in the Oligocene?. In D. R. Prothero, W. A. Berggren [eds.], Eocene-Oligocene climatic and biotic evolution 399-436 Princeton University Press, Princeton, New Jersey, USA.
Levyns M. R.. 1962. Possible antarctic elements in the South African Flora. South African Journal of Science 58: 237-241.
Linder H. P.. 2003. The radiation of the Cape flora, southern Africa. Biological Reviews 78: 597-638.[Medline]
Linder H. P.. 2005. The evolution of diversity: the Cape flora. Trends in Plant Science 10: 536-541.[ISI][Medline]
Linder H. P. Mann D. M.. 1998. The phylogeny and biogeography of Thamnochortus (Restionaceae). Botanical Journal of the Linnean Society 128: 319-357.[CrossRef][ISI]
Linder H. P. Hardy C. R.. 2004. Evolution of the species-rich Cape flora. Philosophical Transactions of the Royal
