| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
0 CSIRO, Plant Industry, Centre for Plant Biodiversity Research, Australian National Herbarium, GPO Box 1600, andThe Australian National University, Division of Botany and Zoology, Canberra, Australia
Received for publication March 15, 1999. Accepted for publication May 25, 1999.
ABSTRACT
The Gnaphalieae are a group of sunflowers that have their greatest diversity in South America, Southern Africa, and Australia. The objective of this study was to reconstruct a phylogeny of the South African Gnaphalieae using sequence data from two noncoding chloroplast DNA sequences, the trnL intron and trnL/trnF intergenic spacer. Included in this investigation are the genera of the Gnaphalieae from the African basal groups, members of the subtribes Cassiniinae, Gnaphaliinae, and Relhaniinae, and African representatives from the large Old World genus Helichrysum. Results indicate that two Gnaphaloid genera, Printzia and Callilepis, should be excluded from the Gnaphalieae. In most trees the Relhaniinae s.s. (sensu stricto) and some of the basal taxa comprise a clade that is sister to the remainder of the tribe Gnaphalieae. The Relhaniinae, which are restricted to Africa, are not a monophyletic group as presently circumscribed, nor are the South African members of Helichrysum, the Cassiniinae and Gnaphaliinae. There is general agreement between our molecular analysis and that of morphology, particularly in the terminal branches of the trees.
Key Words: Asteraceae • Cape Flora • chloroplast DNA • Gnaphalieae • noncoding sequence • phylogeny • Relhaniinae • South Africa
The Gnaphalieae (paper daisies or everlastings) are a group of sunflowers that have their greatest diversity in South America, Southern Africa, and Australia. They are poorly represented in the Northern Hemisphere (e.g., Antennaria Gaertn., Leontopodium R. Brown ex Cass., and Anaphalis A. P. de Candolle). Phylogenetic relationships among the ~187 genera of the Gnaphalieae have been hypothesized through a recent morphology-based cladistic analysis (Anderberg, 1991a
). However, the ubiquitous parallelisms in morphology that exist within the tribe Gnaphalieae, and indeed Asteraceae (Carlquist, 1976
) as a whole, have made it difficult to find conservative (nonhomoplasious) characters that can be used reliably in phylogeny reconstruction. Based on his analysis, Anderberg (1991a)
proposed that the tribe be composed of 15 monophyletic groups or subtribes, which is contrary to traditional taxonomic concepts in the Gnaphalieae as revised by Merxmüller, Leins, and Roessler (1977)
. Merxmüller, Leins, and Roessler (1977)
recognized three subtribes in their Inuleae s.l. (sensu lato) and Anderberg has assembled his 15 taxa (subtribes and/or groups) from two of the subtribes of the Merxmüller, Leins, and Roessler (1977)
classification, namely the Gnaphaliinae and the Athrixiinae.
Of considerable interest within the Gnaphalieae is the phylogenetic position and circumscription of the African subtribe Relhaniinae and its members. Several genera of this subtribe, Oedera L. for example, have historically been placed in other groups (e.g., Kosteletzky, 1833
; Harvey, 1865
; Dyer, 1975
; Anderberg and Källersjö, 1988
). This study, in part, was devised to resolve these questions.
Given the problems of nonhomologous morphological similarities (homoplasies) in the group, we have chosen to explore these relationships with a molecular approach. The objectives of our work were (1) to attempt to reconstruct the phylogeny of the South African Gnaphalieae using sequence data from two relatively short noncoding chloroplast DNA sequences, the trnL intron and trnL/trnF intergenic spacer and (2) to test the monophyly of the Relhaniinae and Helichrysum Mill. and assess their phylogenetic relationships to the basal groups of the Gnaphalieae. This spacer region has proven useful in resolving generic and tribal relationships in the Asteraceae (Bayer and Starr, 1998
). Investigated in this paper are the genera of the Gnaphalieae from the basal African genera, Anisothrix, Athrixia, Arrowsmithia, Callilepis, Pentatrichia, and Printzia, members of the subtribe Relhaniinae, and representatives from the large Old World genus Helichrysum (~500 spp. worldwide). This study is the first part of an ongoing attempt to reconstruct the phylogeny of the Gnaphalieae on a worldwide basis.
Recent analyses of the relationships of the Inuleae, Plucheeae, and Gnaphalieae
The work of Anderberg (1989, 1991a, b, c)
and Karis (1993)
suggests that the tribe Inuleae (Asteraceae: Asteroideae) sensu Merxmüller, Leins, and Roessler (1977)
should be considered as three separate tribal lineages: Inuleae s.s., Gnaphalieae, and Plucheeae. This "three tribe" system, based in part on a cladistic analysis of morphological characters, was later supported by two molecular analyses (Kim and Jansen, 1995
; Bayer and Starr, 1998
), but both studies indicated that the Inuleae s.l. was not a monophyletic lineage. In the trnL intron and trnL/trnF intergenic spacer analysis by Bayer and Starr (1998)
, the Inuleae s.s. and Plucheeae together form a clade sister to the remainder of the Asteroideae. Kim and Jansen (1995)
, using ndhF, also suggest a strong sister relationship of the Plucheeae and Inuleae within the Asteroideae, but the base of their topology was not sufficiently resolved to discern the sister relationships of that clade. The topological relationships identified by Bayer and Starr (1998)
were almost identical to those described by Karis (1993)
based on morphology. Therefore, the segregation of the Gnaphalieae from the Inuleae s.l. was warranted, as they do not form a monophyletic assemblage in any of the abovementioned analyses.
However, the sister relationships of the Gnaphalieae proposed by these studies remains controversial. Bayer and Starr (1998)
, based on the trnL intron and trnL/trnF spacer data, proposed that the Gnaphalieae are sister to a fourth tribe, the Senecioneae. Karis (1993)
, using morphology, revealed them as sister to a clade containing the tribes Astereae and Anthemideae. Jansen, Michaels, and Palmers' (1991)
RFLP (restriction fragment length polymorphism) analysis described the Gnaphalieae as sister to the Inuleae (represented by Inula L.). Keeley and Jansen (1991)
show them as sister to a clade consisting of the Inuleae and Plucheeae. Finally, Kim and Jansen's (1995)
ndhF analysis have the Gnaphalieae in an unresolved clade containing the Calenduleae, Astereae, and Anthemideae.
MATERIALS AND METHODS
Fieldwork
It was established that the greatest diversity could be sampled in the Western Cape Province of South Africa, so fieldwork focused on that region. We collected from as many Gnaphalieae, Senecioneae, and Calenduleae species as possible with a primary goal to collect material from the type species of each genus (Table 1). Leaves for DNA extraction were preserved in liquid CTAB/NaCl solution at ambient temperature and later at -20°C in the laboratory (Rogstad, 1992
).
|
, Bremer (1987, 1994)
, Kim et al. (1992)
, Kim and Jansen (1995)
, and Bayer and Starr (1998)
. Two genera of the Barnadesioideae, Chuquiraga Juss. and Doniophyton Wedd., were chosen as outgroups due to basal position in the Asteraceae of this subfamily in all the abovementioned studies. Their use as an outgroup for the remainder of the Asteraceae is now routine practice (Jansen et al., 1990
; Jansen, Michaels, and Palmer, 1991
; Keeley and Jansen, 1991
; Bayer and Starr, 1998
). Tribal circumscriptions and nomenclature are based on the treatment of the Asteraceae by Bremer (1994)
.
Ingroup sampling of Cichorioideae and Asteroideae
Sequences from representatives of the tribes of the Cichoriodieae and extra-Gnaphaloid members of the Asteroideae were taken from taxa sampled in a previous study of tribal relationships (Bayer and Starr, 1998
). In addition, eight new taxa were sampled to broaden tribal representation, with emphasis on the Asteroideae: Coreopsis tinctoria Nutt. (Heliantheae: Coreopsidinae), Dimorphotheca sinuata DC. (Calenduleae), Euryops virgineus (L. f) Less. (Senecioneae), Pentzia flabelliformis Willd. (Anthemideae), Pluchea dentex R. Br. ex Benth. (Plucheeae), Senecio lineatus DC. (Senecioneae), and S. pterophorus DC. (Senecioneae) (Table 1). Tarchonanthus trilobus DC. (Mutisieae) was also sequenced as it has been recognized as the type genus of a new tribe, Tarchonantheae (Keeley and Jansen, 1991
). Bentham (1873)
included Tarchnanthus in the Inuleae s.l. A broad representation from the extra-Gnaphaloid Asteroideae was used so that the phylogenetic placement of several dubious members of the South African Gnaphalieae might be elucidated.
Except where noted, Anderberg (1991a)
is followed throughout this work as a basis for nomenclature and classification of the Gnaphalieae. One member from each of the available recognized genera of the Gnaphalieae was sequenced, except for the large polymorphic and presumably polyphyletic genus Helichrysum, for which we sequenced 14 species. We followed the classification of Hilliard (1983)
for Helichrysum and selected taxa belonging to 13 of her 30 informal morphological groups. Approximately 70% of the 49 genera of South African Gnaphalieae were sequenced in this study (Table 1). Fifty-three new sequences (72% of total) were generated for this study and have been submitted to GenBank (accession numbers are in Table 1). Voucher specimens for all samples are deposited in the herbaria cited in Table 1.
DNA isolation, amplification, and sequencing
Total DNA was isolated as outlined in Bayer, Hufford, and Soltis (1996)
. Recalcitrant DNAs were purified according to methods outlined in Gilmore, Weston, and Thomson (1993)
, except the amounts of components were scaled down for our purposes. The trnL/F region was amplified via the polymerase chain reaction (PCR) using Taq DNA polymerase. The PCR reaction mixture consisted of 5 µL of 20X reaction buffer, 6 µL of 25 mmol/L magnesium chloride solution, 16 µL of a 1.25 mmol/L dNTP solution in equimolar ratio, 25 pmol of each primer, 10–50 ng of template DNA, and 1.0 unit of polymerase in a total volume of 100 µL. The PCR samples were heated to 94°C for 3 min prior to the addition of DNA polymerase to denature unwanted proteases and nucleases. The double-stranded PCR products were produced via 30 cycles of denaturation (94°C for 1 min), primer annealing (48°C for 1 min), and extension (72°C for 2 min). A 7-min final extension cycle at 72°C followed the 30th cycle to ensure the completion of all novel strands.
The region was usually amplified as a single piece using primers "c" and "f" of Taberlet et al. (1991)
to amplify across the trnL intron and trnL/trnF spacer. In some instances, recalcitrant DNA was amplified as two separate regions using primers "c" with "d" and "e" with "f". Double-stranded PCR products were cleaned by column purification using Wizard® PCR Preps DNA Purification System (Promega Corp., Madison, Wisconsin, USA) prior to sequencing.
The double-stranded PCR products were then used as templates in cycle sequencing reactions, which employed four primers (Taberlet et al., 1991
) to sequence the two regions, including the terminal primers "c" and "f" and, in addition, the internal primers "d" and "e" when the region was amplified as two pieces. The double-stranded PCR products were sequenced using the dideoxy chain termination method (Sanger, Nicklen, and Coulson, 1977
) with the use of the Big Dye Terminator RR Kit® (Perkin-Elmer Applied Biosystems, Norwalk, Connecticut, USA) and an ABI automated sequencer (Perkin-Elmer Applied Biosystems, Norwalk, Connecticut, USA) at CSIRO, Plant Industry. An annealing temperature of 57°C was used for primer "f," while temperatures ranging from 60° to 62°C were employed for primers "c," "d," and "e." The cycle sequencing protocol followed manufacturer's instructions. Sequences were assembled using SequencherTM 3.0 (Gene Codes Corporation, Ann Arbor, Michigan, USA).
The sequences were initially aligned using Clustal V vers. 1.4 (Higgins, Bleasby, and Fuchs, 1992
), then adjusted manually (Swofford and Olsen, 1990
) following the principles of noncoding sequence alignment presented in Kelchner (unpublished data) and Kelchner and Clark (1997)
. To improve homology assessment of the numerous insertions and deletions, gaps were positioned with consideration of mutational mechanisms that likely created the observed length mutations. Indels were placed so as to minimize the number of inferred length mutations unless there was clear evidence that particular length mutation events were nonhomologous. The matrix was surveyed manually for the presence of possible hidden minute inversions (Kelchner and Wendel, 1996
). The matrix is available from R.J.B. on request.
Sequence data analysis
Sequence data were analyzed using PAUP 4.0d64. (Swofford, 1997
) on a MacIntosh G3. The data matrix consisted of two outgroup and 72 ingroup taxa. Phylogenetic reconstruction was performed on unweighted characters by heuristic searches with simple, closest, and furthest addition of taxa. Heuristic searches employing a random addition sequence of 100 replicates were also conducted to search for other islands of most parsimonious trees (Maddison, 1991
). Indels were coded (present/absent) and three separate data sets were analyzed. The first excluded all the coded indels, the second included indel characters only, and the third was the complete matrix including all indels and nucleotide characters. Strict and 50% majority rule consensus trees (Margush and McMorris, 1981
) were constructed for the set of equally most parsimonious cladograms (Fig. 1). Examining the distribution of phylogenetically informative characters (point mutations and indels) on the tree topologies was facilitated by MacClade version 3.0 (Maddison and Maddison, 1992
).
|
) was used to estimate the robustness of clades and were performed using a converse constraint (ENFORCE CONVERSE command) method (Baum, Sytsma, and Hoch, 1994
). In this procedure, multiple heuristic TBR searches using a random addition sequence of 100 replicates were constrained to search for only those trees lacking an hypothesized clade present in the strict consensus tree of parsimony analyses. A simple subtraction of the shortest trees found in these searches from the most parsimonious tree found in parsimony analysis is equal to the Bremer support index for that clade. The amount of phylogenetic information in the parsimony analysis was assessed by use of the consistency index (C.I.; Kluge and Farris, 1969
) and the retention index (R.I.; Farris, 1989
).
Bootstrapping (Felsenstein, 1985
) was considered an inappropriate measure of clade support for this data set due to the violation of its core assumptions by the following aspects of noncoding region molecular evolution (Kelchner, unpublished data): (1) nucleotides do not uniformly evolve independently in intergenic spacers and introns in the chloroplast; (2) mutations are not distributed randomly throughout chloroplast noncoding regions; and (3) the presence of many gaps in the data matrix make removal of indel characters essential, thus severely limiting the number of characters available for bootstrap resampling.
Morphology
MacClade was used to explore the distribution of a broad range of vegetative and floral characters on a tree with similar topology to that found in the molecular analysis. The morphological data matrix presented by Anderberg (1991a)
and Anderberg (1992)
, but consolidated by Puttock (1994)
, was used for this purpose, using only the taxa we included in our analysis. These morphological character scorings, presented in Fig. 3, are predominantly from the matrices of Anderberg (1991a)
, Anderberg (1992)
, Anderberg and Bremer (1991)
; Anderbergia N. Norb. and Langebergia Anderb. were scored by R.J.B. and most other character states for other taxa were verified by R.J.B. Base chromosome numbers (Fig. 3) for each genus are from Anderberg (1991a).
Distributions of genera were compiled from Lundgren (1972, 1974)
, Dyer (1975)
, Bremer (1976, 1978a, b)
, Kroner (1980)
, Hilliard and Burtt (1981)
, Hilliard (1983)
, Anderberg (1988, 1991a)
, and Nordenstam (1996)
, and their distributions are plotted (Fig. 3) against the phytogeographic biomes of Rutherford and Westfall (1994)
.
|
Sequence characteristics
Length variation for the entire trnL intron ranged from a low of 419 nucleotides to a high of 470 nucleotides (Table 2). The proportion of nucleotide differences ranged from 0.0 to 6.7% between all species of Asteraceae (Table 2). The G/C content of the intron averages 35.0%. The complete trnL/F intergenic spacer ranges from 228 nucleotides in Tagetes L. to 402 nucleotides in length in the Asteraceae, but the mean length is 350 nucleotides. The great range in length may be somewhat misleading because Tagetes has a unique 82 nucleotide deletion; the next shortest sequence is 281 nucleotides. The proportion of nucleotide differences in the spacer is greater than that found in the intron ranging from 0.0 to 12.5% between all species of Asteraceae. Similar to the intron, the spacer has an average G/C content of 34.0% (Table 2). Within the Asteraceae, the proportion of nucleotide differences in the combined spacer and intron sequences ranged from 0.0 to 6.9% (Table 2). Total average G/C content is 34.5%. A total of 147 sites (14.0% of the sequence length) potentially provide phylogenetic information, but the other sites (86.0%) are either invariant or are strictly autapomorphic. Thirty-four indels (Table 2), ranging in length from one to 27 nucleotides, could be coded unambiguously for inclusion in the phylogenetic analysis. The spacer contains more phylogenetically informative indels than the intron does, although the ratios of potentially informative indels to potentially informative nucleotides (1:4.1 and 1:4.7, respectively) are very similar for the intron and spacer.
|
) on the data sets did not reveal any islands of trees of shorter length. Also presented is one of the 366 most parsimonious trees illustrating that the Gnaphalieae s.l., as presently circumscribed, may not be monophyletic (Fig. 2). A reduced molecular tree, with all taxa for which morphological data are available, is presented in Fig. 3.
|
, in most cases tribes that were represented by more than one genus [i.e., the Anthemideae (clade M), Calenduleae (clade J), Helianthieae s. l. (clade G), Plucheeae (Clade O), Lactuceae (clade N), Senecioneae (clade I)] are monophyletic. The only exception to this is the Cardueae in the Cardueae/Mutisieae clade (K). Tarchonanthus L. is part of the Cardueae/Mutisieae clade (K) lending support to Bremer's (1994)
placement of the genus in the Mutisieae.
Generally, tribal support throughout the topology (Fig. 1) is high, with most clades being supported by multiple synapomorphies, from three in the Helianthieae s. l. (G) to 20 in the Anthemideae (clade M). The Gnaphalieae, in the sense of Anderberg (1991a)
, are not monophyletic in this analysis because the genera Callilepis DC. and Printzia Cass. are associated with the Astereae (Fig. 1; clade H). Although Callilepis is not in clade H with Aster L. and Printzia (SYN = 6), it does share an insertion with Aster (Fig. 1, arrow). Additionally, in several shortest trees the basal group of the Gnaphalieae (Figs. 1, 2; clade F) becomes sister to other tribes, such as the Astereae, Anthemideae, and Heliantheae (Fig. 2; clade G, H). Clade F contains members of the Relhaniinae s.s. and some of Anderberg's (1991a) "basal"
group of the Gnaphalieae. The Gnaphalieae in a broad sense (clade Q) are not well supported (SYN = 2), as the group does not appear in all most parsimonious trees.
Topology of clades within the Gnaphalieae s. l. (clade Q)
The following considers the Gnaphalieae in a broad sense (clade Q) to include the members of the Relhaniinae and some of Anderberg's "basal" group (clade F), and the Gnaphalieae s.s. as clade P (Fig. 1). Beginning at the base of the group (clade F) and proceeding systematically toward the top of the tree (clade T) a number of well-supported groups are evident. The basal group of the Gnaphalieae as represented in this anlayis is clade F, which contains three taxa from Anderberg's "basal" group (BG), Pentatrichia Klatt, Anisothrix O. Hoffm., and Arrowsmithia DC. and the Relhaniinae s.s. (R). Pentatrichia and Anisothrix are sister taxa and together form a clade that is sister to the strongly supported (SYN = 4; BSI = 6) clade F' in which Arrowsmithia is sister to the Relhaniinae s.s. A strong relationship of Leysera L. to Rhynchopsidium DC and A. DC. (SYN = 5; BSI = 6) is evident, and they are sister to a clade containing Rosenia Thunberg, Relhania L'Her, and Oedera.
The Gnaphalieae s.s. (clade P) are strongly supported (SYN = 5; BSI = 4) with Trichogyne Less. sister to the rest of the group (Fig. 1). Next is clade E containing shrubby members of the Relhaniinae s. l., Metalasia R. Br., Phaenocoma D. Don, Lachnospermum Willd., and Dolichothrix Hilliard and B. L. Burtt as well as with Athrixia Ker-Gawl. Athrixia is the only disparate member of this clade, having been placed as part of the basal group in Anderberg's (1991a)
analysis (Fig. 1). Lasiopogon Cass., a member of Anderberg's subtribe Gnaphaliinae, is sister to clade U, showing more sequence affinites to this clade than to other members of the Gnaphaliinae, such as Vellereophyton Hilliard and B. L. Burtt (Fig. 1). Clade D (SYN = 1; BSI = 2) contains four genera of ericoid-leaved shrubs, Bryomorphe Harv., Disparago Gaertn., Elytropappus Cass., and Stoebe L., which are currently members of the Relhaniinae of Anderberg (1991a)
. Vellereophyton is part of a polytomy that comprises other members of the Gnaphaliinae of clade C, including Plecostachys Hilliard and B. L. Burtt and Tenrhynea Hilliard and B. L. Burtt. Also part of this polytomy is a large clade (R) in which Oreoleysera K. Bremer, currently in subtribe Relhaniinae, is sister to the last two Clades A and B, consisting primarily of members of the Cassiniinae and Gnaphaliinae, respectively. Clade B contains four members of Anderberg's Cassiniinae, Petalacte D. Don, Langebergia, Anderbergia, and Anaxeton Gaertn., as well as Syncarpha DC. and two species of Helichrysum, H. cylindriflorum (L.) Hilliard and Burtt and H. dasyanthum (Willd.) Sweet, all three of the subtribe Gnaphaliinae (Fig. 1). In clade A, Edmondia Cass., which was previously part of both Helichrysum and Helipterum DC. (Hilliard, 1983
; Nordenstam, 1989
), is sister to a well-supported lineage (SYN = 4; BSI = 5) of 12 South African Helichrysum species representing 11 of Hilliard's informal groups (Hilliard, 1983
).
DISCUSSION
The most diverse family of flowering plants in Southern Africa are the Asteraceae, with 174 genera, 80 of which are endemic, and 2072 endemic species (Goldblatt, 1978
). Most of the genera of the South African Gnaphalieae are confined to or have their greatest diversity in the Cape floristic region (Fig. 3). The fact that Anderberg's (1991a)
basal group of the Gnaphalieae is restricted in distribution to southern Africa suggests the original radiation of the group took place in South Africa.
Anderberg's (1991a)
"basal" group of the Gnaphalieae and the Relhaniinae consists primarily of genera that Merxmüller, Leins, and Roessler (1977)
had placed in the subtribe Athrixiinae of their Inuleae. Merxmüller, Leins, and Roessler (1977)
were "disturbed" by this subtribe, realizing it was not well defined and had a number of disparate members that did not seem to fit in the group, or any other group for that matter. One of the most anomalous of these genera is Callilepis, and indeed, the phylogenetic relationships proposed in Fig. 1 indicate Callilepis is not included in the Gnaphalieae (clade Q).
The affinities of Callilepis are not clear from our analysis, as it is part of a polytomy in the majority rule topology (Fig. 1). However, it does share a single base pair deletion with Aster (Fig. 1, arrow), which could point to an affinity there. Its heads are radiate, the pappus consists of a few awns, and the style branches have stigmatic hairs localized near the apices, characteristics that are unlike most of the Gnaphalieae. It shares an unusual chemical character with Atractylis L. (Cardueae) (Candy et al., 1977
), but that may be convergence as most evidence points to Asteroideae affinities for Callilepis, not Cichorioideae. In another chemical investigation Bohlmann and Zdero (1982)
discovered additional compounds in Callilepis indicating no clear relationships of Callilepis to members of the Athrixiinae sensu Merxmüller, Leins, and Roessler (1977)
. As we continue to add sequence data to our data set the phylogenetic position of Callilepis will hopefully be clarified; however, at present our evidence indicates Callilepis is probably not a member of the Gnaphalieae.
Another taxon conflicting with Anderberg's (1991) Gnaphalieae is Printzia. On seeing Printzia for the first time in the field, we (R.J.B. and C.F.P.) were impressed by its resemblance to species of Aster, and interestingly the type species of Printzia, P. polifolia (L.) Hutch. (Table 1), was originally described by Linnaeus as Aster polifolius L. (Kroner, 1980
). In our analysis, Printzia and Aster form a monophyletic group supported by six synapomorphies in 97% of the most parsimonious trees. Additionally, in a study of the phytochemistry of Printzia, Bohlmann and Zdero (1978)
found compounds that were identical to those found in members of the Astereae, specifically to species of Solidago L. Given the evidence, we feel the affinities of Printzia lie with the Astereae, not the Gnaphalieae.
The Relhaniinae s.s. and some "basal" group taxa [the core Athrixiinae of Merxmüller, Leins, and Roessler (1977)
] comprise clade F. In the majority of most parsimonious trees, this group is sister to the remainder of the tribe Gnaphalieae (Fig. 1), but in some it is sister to a clade containing the Anthemideae and Helianthieae s. l. (Fig. 2; clades G, H). It is interesting to note that historically Oedera has most often been considered as a member of the Anthemideae (Anderberg and Källersjö, 1988
), although it was originally placed in the Inuleae by Linnaeus (1771). The primary reason for lodging Oedera with the Anthemideae is the lack of tailed anthers, truncate and apically penicillate style branches, and pappus of scales (Anderberg and Källersjö, 1988
). These are features that are shared among other members of clade F (Fig. 3), but some are relatively uncommon in the Gnaphalieae s.s. A similarity among Oedera and Leysera and Relhania was noted by Anderberg and Källersjö (1988)
, who argued that removal of Oedera to the Inuleae and later Gnaphalieae (Anderberg, 1991a
) was better than moving Leysera and Relhania to the Anthemideae. At this time, there appear to be two alternative hypotheses as to the relationships of this group, either it is the basal group of the Gnaphalieae or it might be a separate tribe itself, i.e., the Relhanieae Kostel., as has been proposed in the distant past by Kosteletzky (1833).
The Relhaniinae s.s. do not appear to be part of the Anthemideae, but may be sister to it (Fig. 2). There are a number of morphological features (Fig. 3) that hold the Relhaniinae s.s. itself rudimentarily together, including frutescent growth habit, a pappus of scales, five vascular bundles in the cypsela wall, and paleate receptacles. Chromosome numbers (Fig. 3) in the group appear to be based on x = 7, with aneuploid increase in Leysera to x = 8 and a decrease to x = 5 in the annual genus Rhynchopsidium. In clade F'' we have Pentatrichia and Anisothrix as sister taxa. Anderberg (1988, 1991a)
alluded to the morphological similarity of these two taxa, and they are unusual among the Gnaphalieae for having leaves mostly with dentate margins [excepting A. integra (Compton) Anderb.]. The Gnaphalieae almost exclusively have leaves with entire margins. Anisothrix has some of the rare leysseral derivative compounds also found in Relhania, Leysera, and Macowania Oliv. (Zdero, Bohlmann, and Anderberg, 1991
).
Within the Relhaniinae s.s., Arrowsmithia is sister to the traditionally (Anderberg and Bremer, 1991
) recognized members of the Relhania group. It shares a number of morphological features with the Relhaniinae, namely radiate heads, paleate receptacles, and frutescent habit (Fig. 3), and is also strongly supported as a member of the group based on our molecular data (Fig. 1). The brownish color, texture, and size of the phyllaries of Arrowsmithia are very similar to members of the Relhaniinae (R.J.B., personal observations). Arrowsmithia is a monotypic genus that has been included within Macowania (Kroner, 1980
), and the close relationship of these taxa has been recognized by others (Hilliard and Burtt, 1985
; Anderberg, 1991a
). A chemical investigation of Macowania (Bohlmann and Zdero, 1977
) showed that two compounds isolated from this genus were also found in Leysera gnaphalodes L. Harvey (1865) was perhaps the first to note the similarity in "aspect" between Arrowsmithia and Relhania. Although Anderberg (1991a)
had Arrowsmithia as part of his basal group, quite removed from the Relhaniinae, our current results, the morphological characteristics mentioned above, and the phytochemical evidence of Bohlmann and Zdero (1977)
indicate a probable sister relationship to the Relhaniinae s.s.
Within the Relhaniinae s.s. (Fig. 1) we find a strong sister relationship between Leysera and Rhynchopsidium, a relationship also indicated by the morphological cladistic analysis of Anderberg and Bremer (1991)
. Rhynchopsidium had once been part of Leysera (Bremer, 1978a
), but was removed from Leysera based on a number of apomorphies, including annual habit. In the remaining clade we find Oedera as sister to Relhania and they in turn sister to Rosenia (Fig. 1; clade F'). A similar relationship was demonstrated by Anderberg and Bremer (1991)
; in their analysis Rosenia is somewhat more removed from Oedera and Relhania.
The only member of the Relhaniinae s.s. that is not part of the group in clade F' in our analysis is the monotypic genus Oreoleysera. Instead, Oreoleysera is in a monotypic clade that is sister to clade R, containing Helichrysum species and elements of the South African Cassininae (Fig. 1). Bremer (1978b)
removed Leysera montana Bolus from the genus Leysera and placed in its own genus, Oreoleysera, because he felt the genus was nonmonophyletic when the species was included. Oreoleysera montana (Bolus) K. Bremer is an alpine cushion plant from the Cape Region and as such has highly reduced and modified morphology. Leysera has a pappus of dentoid scales (Fig. 3) and plumose pappus bristles, while Oreoleysera has small narrow scales and nonplumose bristles. There has been some question as to the homology of those scales, with Bremer (1978b)
indicating they are absent, or at least are similar to those seen in Leysera. Later, Anderberg and Bremer (1991)
reversed the earlier statement, claiming they are present. Additionally, Oreoleysera is epaleate, whereas all the Relhaniinae s.s. (including Arrowsmithia; Fig. 3, clade F') are paleate. The analyses of Anderberg (1991a)
and Puttock (1994)
show Oreoleysera as sister to the rest of the Relhaniinae s.s. Oreoleysera has probably been misplaced in the Relhaniinae due to its reduced morphology and alpine habit convergence.
Trichogyne was incorporated into the genus Ifloga Cass. by Hilliard and Burtt (1981)
, but Anderberg (1991a)
disagreed and reinstated Trichogyne to represent eight species in South Africa, leaving six other South African and Saharo-Sindian species in Ifloga. Anderberg's cladistic analysis of morphology (Anderberg, 1991a
) shows Ifloga and Trichogyne in a strongly supported sister relationship. Traditionally this group has been placed in the subtribe Filaginae of the Inuleae s.l. (Bentham, 1873
), but Merxmüller, Leins, and Roessler (1977)
put it in the Gnaphaliinae (Gnaphalium L.—Helichrysum group). Anderberg (1991a)
follows the lead of Merxmüller, Leins, and Roessler (1977)
, also positioning the group in his Gnaphaliinae. Hilliard and Burtt (1981) comment that Ifloga and Trichogyne remain isolated members of the Gnaphaliinae, implying the Filaginae affiliation is more convincing. Our analysis shows Trichogyne as sister to clade E, a group that shares the apomorphies of shrubby habit, small ericoid leaves, small few-flowered heads (as in Metalasia), one row of barbellate–plumose pappus bristles, and epaleate receptacles (Fig. 3). Therefore, we believe the affinities of Trichogyne may lie with the members of clade E.
In clades E and D (Fig. 1) are the members of Anderberg's Metalasia group that we surveyed. These are the remaining elements of Anderberg's (1991a)
Relhaniinae s.l. The only enigmatic member of either clade is Athrixia, of Anderberg's basal group, in clade E (Fig. 1). It differs from other members of clade E by having radiate heads and nonericoid leaves, but radiate heads do appear in Bryomorphe and Disparago in the other Metalasia group, clade D (Fig. 3). Athrixia aside, the two clades (D and E) of Anderberg's (1991a)
Metalasia group share a number of apomorphies (Fig. 3), including shrubby habit, ericoid leaves, five vascular bundles in the cypsela wall, pericyclic cambium, and possibly a base chromosome number of x = 8. All members of clade E except Lachnospermum share papery, opaque, dimorphically colored phyllaries that are a color other than hyaline or brown, as well as pappus bristles that are barbellate or scabrous. The members of clade D, in contrast, all have plumose pappus bristles (except Bryomorphe, which is barbellate) and cartilaginous, transparent, monomorphically colored involucral bracts that are hyaline or brown in color. Even though Anderberg's Metalasia group as a whole seems fairly cohesive morphologically, we have found it paraphyletic or polyphyletic as it does not form a clade on the topology (Figs. 1 and 3). Phaenocoma and Stoebe share a similar phytochemistry (Bohlmann and Suwita, 1978
), further uniting Anderberg's (1991) Metalasia group. The taxa included within clades D and E (Fig. 3) correspond to Anderberg's (1991a)
groups based on morphology, with the exception that Bryomorphe was sister to the other two clades of the Metalasia group in his study. Bryomorphe, like Oreoleysera, is an alpine cushion plant; subsequently it has highly reduced and convergent morphology presumably driven by the alpine environment, hence its phylogenetic position may have been obscured by homoplasious morphological characters. Lasiopogon is in a rather unusual position, allied to the two Metalasia clades, but sharing few morphological apomorphies with them (Fig. 3). It is a small annual in subtribe Gnaphaliinae and is placed close to Trichogyne by Anderberg.
On the whole, in clade S we see a trend, except in Oreoleysera, toward the following charater states: non-ericoid leaves, 2 (1–3) vascular bundles in the cypsela wall, cypselae with trichomes, absence of pericyclic cambium, and perhaps a tendency toward a base chromosome number of x = 7 (Fig. 3). Plecostachys and Tenrhynea are sister taxa in our analysis (Fig. 1) and in Anderberg's (1991a)
. They share a number morphological apomorphies with Vellereophyton, such as lack of distinctly tailed anthers and herbaceous habit (Fig. 3). All three belong to Anderberg's (1991a)
subtribe Gnaphaliinae, and both Plecostachys and Vellereophyton are recent segregates out of the large, polyphyletic genus Gnaphalium. Anderberg's cladistic analysis of morphology shows the three genera as being fairly closely related and this conclusion is supported by our molecular analysis.
Clade B consists of all the South African members of Anderberg's (1991a)
Cassiniinae (Anaphalis group), as well as a few members of the Gnaphaliinae, Syncarpha, and two species of Helichrysum s. l. Anaxeton, Petalacte, Langebergia, and Anderbergia represent a cohesive group on morphological grounds in Anderberg's (1991a)
analysis. In fact, Langebergia and Anderbergia were recently created segregate genera from former Petalacte species (Anderberg, 1991a
; Nordenstam, 1996
). The strong sister relationship of Langebergia and Anderbergia cannot be denied, based on both morphological and molecular grounds (Fig. 1; SYN = 3, BSI = 3, and they share two large nonhomoplasious insertions).
Syncarpha, which was resurrected by Nordenstam (1989)
to accommodate misaligned species of Helipterum and Helichrysum, has similarities to the Anaphalis group. They all have monochromous, opaque phyllaries, epaleate receptacles (except Petalacte), and barbellate (infrequently plumose) pappus. Anderberg's analysis (1991a)
did not test the relationships between the Anaphalis group and Syncarpha, but Puttock (1994)
showed them to be nested in a polytomy of members of the Gnaphaliinae and sister to a clade containing Syncarpha. Perhaps Syncarpha lies within the Anaphalis group in the Cassiniinae rather than with the Gnaphaliinae; however, its affinities remain unclear until more species can be evaluated.
The presence of the two species of Helichrysum in clade B is supportive of the idea that Helichrysum is a polyphyletic genus. Helichrysum dasyanthum is sister to Anaxeton, and they in turn are sister to Syncarpha. An usual synapomorphy it shares with Syncarpha is that of fimbrilliferous receptacles, which are rare in the Gnaphalieae, appearing only in Syncarpha, Edmondia, and some species of Helichrysum [absent from H. orientale (L.) Gaertn., the type of the genus]. Helichrysum dasyanthum bears a morphological resemblance to Anaxeton in that it is a small shrub with leaves up to 30 mm long, the heads are born in terminal corymbs and are heterogamous, and the phyllaries are opaque to semi-opaque (Hilliard, 1983
). The pappus bristles are barbellate, and the florets, especially the pistillate ones, are covered with glandular hairs near the apex. Lundgren (1972)
commented on the problems of generic delimitation in the Gnaphalieae, but it was his opinion that Anaxeton was a comparatively well-defined genus. However, it may be that the generic concept of Anaxeton could be expanded to include H. dasyanthum and other Helichrysum species.
The affinities of Helichrysum cylindriflorum are less obvious. It is a small shrub with large numbers of tiny heads arranged in corymbs or panicles, the phyllaries are semi-opaque, the pappus is of barbellate bristles, the corolla is glandular at the apex, and the receptacle is epaleate. Since it is part of a polytomy in this analysis, its relationships within clade B are difficult to assess. Perhaps further sampling of taxa within this clade could provide resolution to this problem.
In clade A, Edmondia is sister to a well-supported clade of 12 Helichrysum species representing 11 of Hilliard's informal groups (Hilliard, 1983
). Helichrysum and Edmondia seem quite distinct genera with a large number of morphological differences. The leaves of Edmondia are concave to involute and pubescent abaxially only, while in Helichrysum they are flat and pubescent on both surfaces. In Edmondia the capitula are solitary or few, whereas in Helichrysum they are many and generally in flat-topped corymbs or panicles. The pappus of Edmondia is dimorphic and the bristles cohere in a ring, whereas in Helichrysum the bristles are monomorphic and freely separate from each other when the cypselae are dispersed. Edmondia is a lineage that is strongly differentiated from the remainder of Helichrysum s.s.
The lineage containing the majority of the Helichrysum species (clade T) is Helichrysum s.s., at least until we can determine the affinities of the type of the genus, H. orientale. Out of the 30 groups of Helichrysum recognized by Hilliard (1983)
, we collected and sampled members of 13 of these groups. The apparent polyphyly of Helichrysum in this study illustrates the need for more sequencing and evaluation of the genus. In a large, polymorphic genus such as Helichrysum it is often difficult to find synapomorphies that will unite all the taxa. However, the combination of characters used to circumscribe the genus are smooth, honey-combed, or fimbrilliferous receptacles, glandular hairs on the abaxial surface of the corolla lobes, anthers generally possessing apical appendages, style branches truncate or penicillate, cypselae generally glabrous or with small hairs, pappus usually of scabrous, barbellate, or plumose bristles, and bases free or coherent only by the presence of patent cilia (Hilliard, 1983
). It is difficult to assess relationships of taxa within this clade, as we need to sequence many more species before we can draw any meaningful conclusions. However, there are some interesting trends within this clade that are worthy of discussion. The basal group consists of H. acrophilum Bolus (Hilliard's group 17), H. leontonyx DC (group 4), and H. zwartsbergense Bolus. (group 4). It is also interesting to note that both H. leontonyx and H. zwartsbergense belong to the same group (group 4) of Hilliard (1983)
. This is the only case where we have included two species from the same morphological group, and they are sister taxa in our analysis. Both are small-leaved annuals or weak perennials, with small heads bearing phyllaries that do not radiate. The next resolved clade of Helichrysum contains H. tricostatum (Thunb.) Less. (group 11), H. aureum (Houtt.) Merrill (group 30), and H. argyrophyllum DC. (group 29). Similarities between H. argyrophyllum and the other two species are not obvious, but there are some resemblances between H. tricostatum and H. aureum. Both of these species have large showy, yellow, radiating phyllaries and heterogamous heads, receptacles that are shortly honey-combed, pistillate flowers with a conspicuous limb, and glabrous cypselae. The similarity among H. allioides Less., H. dregeanum Sond. and Harv., and H. patulum (L.) D. Don is also not immediately obvious, but both H. allioides and H. patulum have fimbrilliferous receptacles. On the other hand, both H. dregeanum and H. patulum are small-leaved perennials with numerous small heads and scabrid pappus bristles joined by patent cilia. Helichrysum populifolium DC is a taxonomically isolated species (Hilliard, 1983
) with its large cordate leaves and large, open, paniculate synflorescences. Helichrysum cephaloideum Less. and H. odoratissimum (L.) Sweet share a number of apomorphies such as heads in compact corymbose, to almost globose, panicles, yellow phyllaries, and fimbrilliferous receptacles.
A summary of the general morphological trends in the Gnaphalieae correlating with the molecular tree in Fig. 3 follows. The basal group (clade F) consists mostly of shrubs with radiate heads, paleate receptacles, pappus of scales, and untailed anthers. The next group of major clades (D and E) are mostly shrubs with nonradiate heads, pericyclic cambia, ericoid leaves, epaleate receptacles and distinctly tailed anthers. The final clade (S) represents mostly herbaceous taxa that have nonericoid leaves, nonradiate heads, 2–3 vascular bundles in their cypsela walls, pappus of bristles, epaleate receptacles, nonpericyclic cambium, and untailed anthers. There is general agreement of taxon relationships between our molecular analysis and that based on morphology by Anderberg (1991a)
, with respect to Anderberg's small informal groups within his subtribes. However, our analysis differs from Anderberg's in that the Relhaniinae are not supported as a monophyletic group, nor is there strong evidence for the overall subtribal classification of Anderberg (1991a)
. The subtribes will no doubt need further realignment.
Phytogeography of the Gnaphalieae in South Africa
Anderberg's phylogeny (1991a)
indicates that the basal members of the Gnaphalieae belong to genera that are confined to Africa, particularly South Africa. Historical biogeography would therefore suggest that South Africa is the place of origin of the tribe, followed by radiation to other continents, especially Australia and South America. Since we have not included any extra-African members of the Gnaphalieae in our analysis, it is difficult to corroborate this scenario. However, we can summarize its general phytogeographic patterns in South Africa.
Massive speciation events in the Cape Flora likely began during the Pliocene (Goldblatt, 1978
) beginning ~7 million years before present. The Cape Flora probably has its principal origins from temperate and tropical Africa (Goldblatt, 1978
) and secondarily from Eurasia. It is thought that very few of the taxa had Austral origins from South America or Australia (Goldblatt, 1978
), as scant genera are shared between the Cape and these two continents.
Of the genera of Gnaphalieae s.l. we surveyed, almost all are endemic to southern Africa (except Helichrysum, Lasiopogon, Leysera, and Stoebe), and a large proportion of these are restricted to the Cape Flora (Fig. 3). This is not surprising as the Cape Flora is renowned for its species richness and high degree of endemism (95.8% of the species are endemic; Wiemarck, 1941; Oliver, Linder, and Rourke, 1983
). This high species richness and endemism have been linked to its unusual infertile soils, extreme climatic fluctuations in the past, and its current Mediterranean climate with a strong winter rainfall pattern (Goldblatt, 1978
). A few genera, including Pentatrichia, Lasiopogon, and Tenrhynea, occur only in areas outside the Cape Flora (Fig. 3).
The Relhaniinae s.s. (clade F) are most diverse in the Cape region, except for the genus Pentatrichia of Namibia and northwestern South Africa, and Arrowsmithia of the Afromontane forest of the Eastern Cape. It is only the most diverse genera that have extended their ranges beyond the Cape; those that are restricted entirely to the Cape region tend to be less speciose or monotypic genera, such as Phaenocoma and Anisothrix. In clade P, Trichogyne is sister to the rest of the Gnaphalieae s.s. and it has a distribution that is restricted to the Cape, karoo, and desert biomes of western South Africa (Fig. 3). In clades D and E, the Metalasia group consists of genera that are most diverse in the Cape region, but many also occur outside the Cape, where they are represented often by one or a few widespread species. Stoebe has extended its range to the Mascarene Islands, but several other genera such as Dolichothrix, Bryomorphe, and Phaenocoma are narrow endemics. Clade C, containing Plecostachys and Tenrhynea, represents a group that is distributed around the coast of South Africa from the Cape through to the subtropical lowland forest and savanna biomes of Kwazulu Natal. Oreoleysera and its sister clade, clade B, including Helichrysum dasyanthum and H. cylindriflorum,, are restricted to the Cape region and perhaps represent a recent radiation in the South African Gnaphalieae lineage, as they represent the most derived group on our topology. The clade containing Edmondia and Helichrysum s.s. (clade A) is unique for in its great diversity. Edmondia is a genus of three species endemic to the Cape region, but Helichrysum is found across South Africa and has ~250 species there. This has likely been the result of the possible rapid and recent radiation of Helichrysum, both in the Cape region and in the mountains of Kwazulu Natal, the Orange Free State, Eastern Cape and Lesotho. The genus is particularly diverse in the alpine (afroalpine zone) of these eastern mountains and is the predominant genus of Gnaphalieae there (Fig. 3). Unfortunately, it is difficult to discuss anything about the phytogeographic origins of Helichrysum until the recircumscription of this genus is complete.
The distribution of the South African Gnaphalieae discussed above seems to indicate that this tribe radiated in the Cape Flora region. Perhaps the expansion and contraction of the fynbos and karoo created islands of diversity leading to the formation of endemic taxa, as proposed by Anderberg (1991a)
. Ongoing investigations should provide further information about the origin, distribution, and diversity of the Gnaphalieae.
FOOTNOTES
1 The authors thank Ted Oliver (NBG) for supplying leaf material of Disparago ericoides and Peter Goldblatt (MO) for Rosenia glandulosa used in this study, Peter Linder (BOL) and John Rourke (NBG) for logistical assistance for R.J.B. and C.F.P. in the field in South Africa, Josephine Beyers (NBG) for help in determination of material collected, Ann Langston (CANB) for assistance in the lab, Peter Linder for information he supplied to us on the phytogeography of South Africa, and Mark Clements, Mike Crisp, Mike Dillon, Rod King, Rogier de Kok, and Judy West for suggested improvements to our manuscript. This research was supported by a National Geographic Grant to R.J.B. and C.F.P. 
2 Author for correspondence. (e-mail: r.bayer{at}pi.csiro.au
).
3 Current address: Herbarium Pacificum, Department of Natural Sciences, Bishop Museum, 1525 Bernice Street, Honolulu, Hawaii, 96817-2704 USA.
LITERATURE CITED
Anderberg, A. A. 1988 The genus Anisothrix O. Hoffm. (Compositae—Inuleae). Botanische Jahrbücher für Systematik, Pflanzengeschichte und Pflanzengeographie 109: 363–372.
———. 1989 Phylogeny and reclassification of the tribe Inuleae (Asteraceae). Canadian Journal of Botany 67: 2277–2296.
———. 1991a Taxonomy and phylogeny of the tribe Gnaphalieae (Asteraceae). Opera Botanica 104: 1–195.
———. 1991b Taxonomy and phylogeny of the tribe Inuleae (Asteraceae). Plant Systematics and Evolution 176: 75–123. [CrossRef][ISI]
———. 1991c Taxonomy and phylogeny of the tribe Plucheeae (Asteraceae). Plant Systematics and Evolution 176: 145–177. [CrossRef][ISI]
———. 1992 Cladistics of the Gnaphalieae, additional data. Compositae Newsletter 20/21: 35.
———, and K. Bremer. 1991 Parsimony analysis and cladistic reclassification of the Relhania generic group (Asteraceae—Gnaphalieae). Annals of the Missouri Botanical Garden 78: 1061–1072. [CrossRef][ISI]
———, and M. Källersjö. 1988 The tribal position of Oedera L. (Compositae). Botanical Journal of the Linnaean Society 96: 323–332.
Baum, D. A., K. J. Sytsma, and P. C. Hoch. 1994 A phylogenetic analysis of Epilobium (Onagraceae) based on nuclear ribosomal DNA sequences. Systematic Botany 19: 363–388. [CrossRef][ISI]
Bayer, R. J., L. Hufford, and D. E. Soltis. 1996 Phylogenetic relationships in Sarraceniaceae based on rbcL and ITS sequences. Systematic Botany 21: 121–134.
———, and J. R. Starr. 1998 Tribal phylogeny of the Asteraceae based on two non-coding chloroplast sequences, the trnL intron and trnL/trnF intergenic spacer. Annals of the Missouri Botanical Garden 85: 242–256. [CrossRef][ISI]
Bentham, G. 1873 Compositae. In G. Bentham and J. D. Hooker, Genera plantarum 2: 163–533.
Bohlmann, F., and A. Suwita. 1978 Neue Phloroglucin-Derivate aus Leontonyx-Arten sowie weitere Verbindungen aus Vertretern der Tribus Inuleae. Phytochemistry 17: 1929–1934. [CrossRef][ISI]
———, and C. Zdero. 1977 Ein neues Clerodan-derivat sowie weitere Inhaltsstoffe aus der Gattung Macowania. Phytochemistry 16: 1583–1586.
———, and ———. 1978 Diterpenes with a new carbon skeleton from Printzia laxa. Phytochemistry 17: 487–489.
———, and ———. 1982 Five tricyclic sesquiterpenes from Callilepis salicifolia. Phytochemistry 21: 139–142.
Bremer, K. 1976 The genus Rosenia (Compositae). Botaniska Notiser 129: 97–111. [ISI]
———. 1978a The genus Leysera (Compositae). Botaniska Notiser 131: 369–383. [ISI]
———. 1978b Oreoleysera and Antithrixia, new and old South African genera of the Compositae. Botaniska Notiser 131: 449–453. [ISI]
———. 1987 Tribal interrelationships of the Asteraceae. Cladistics 3: 210–253.
———. 1988 The limits of amino acid sequence data in angiosperm phylogenetic reconstruction. Evolution 42: 795–803. [CrossRef][ISI]
———. 1994 Asteraceae: Cladistics and classification. Timber Press, Portland, Oregon, USA.
Candy, H., K. Pegel, B. Brookes, and M. Rodwell. 1977 The occurrence of atractyloside in Callilepis laureola. Phytochemistry 16: 1308–1309.
Carlquist, S. 1976 Tribal interrelationships and phylogeny of the Asteraceae. Aliso 8: 465–492.
Dyer, R. A. 1975 The genera of southern African flowering plants, 1. Department of Agricultural Technical Services, Pretoria, South Africa.
Farris, J. S. 1989 The retention index and the rescaled consistency index. Cladistics 5: 417–419. [ISI]
Felsenstein, J. 1985 Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39: 783–791. [CrossRef][ISI]
Gilmore, S., P. H. Weston, and J. A. Thomson. 1993 A simple, rapid, inexpensive, and widely applicable technique for purifying plant DNA Australian Systematic Botany 6: 139–148.
Goldblatt, P. 1978 An analysis of the flora of southern Africa: Its characteristics, relationships, and origins. Annals of the Missouri Botanical Garden 65: 369–436. [CrossRef][ISI]
Harvey, W. H. 1865 Compositae. In W. H. Harvey, and O. W. Sonder: Flora Capensis: Being a systematic description of the plants of the Cape Colony, Caffraria, and Port, Natal. vol. 3. L. Reeve and Co., Ashford, Kent, UK.
Higgins, D. G., A. J. Bleasby, and R. Fuchs. 1992 CLUSTAL V: improved software for multiple sequence alignment. Computer Applications in the Biosciences 8: 189–191.
Hilliard, O. 1983 Flora of Southern Africa, Part 7 Inuleae, Fascicle 2 Gnaphaliinae. Government Printer, Pretoria, South Africa.
———, and B. L. Burtt. 1981 A revision of Ifloga in southern Africa, with comments on northern hemisphere species. Botanical Journal of the Linnaean Society 82: 293–312.
———, and ———. 1985 Notes on some plants of southern Africa chiefly from Natal: 11. Notes of the Royal Botanic Garden, Edinburgh 42: 227–260.
Jansen, R. K., and J. D. Palmer. 1987 A chloroplast DNA inversion marks an ancient evolutionary split in the sunflower family (Asteraceae). Proceedings of the National Academy of Sciences, USA 84: 1–5.
———, and ———. 1988 Phylogenetic implications of chloroplast DNA restriction site variation in the Mutisieae (Asteraceae). American Journal of Botany 75: 753–766. [CrossRef][ISI]
———, K. E. Holsinger, H. J. Michaels, and J. D. Palmer. 1990 Phylogenetic analysis of chloroplast DNA restriction site data at higher taxonomic levels—an example from the Asteraceae. Evolution 44: 2089–2105. [CrossRef][ISI]
———, H. J. Michaels, and J. D. Palmer. 1991 Phylogeny and character evolution in the Asteraceae based on chloroplast DNA restriction site mapping. Systematic Botany 16: 98–115. [CrossRef][ISI]
Karis, P. O. 1993 Morphological phylogenetics of the Asteraceae-Asteroideae, with notes on character evolution. Plant Systematics and Evolution 186: 69–93. [CrossRef][ISI]
Keeley, S. C., and R. K. Jansen. 1991 Evidence from chloroplast DNA for the recognition of a new tribe, the Tarchonantheae, and the tribal placement of Pluchea (Asteraceae). Systematic Botany 16: 173–181. [CrossRef][ISI]
Kelchner, S. A., and L. G. Clark. 1997 Molecular evolution and phylogenetic utility of the chloroplast rpl16 intron in Chusquea and the Bambusoideae (Poaceae). Molecular Phylogenetics and Evolution 8: 385–397. [CrossRef][ISI][Medline]
———, and J. F. Wendel. 1996 Hairpins create minute inversions in non-coding regions of chloroplast DNA. Current Genetics 30: 259–262. [CrossRef][ISI][Medline]
Kim, K-J., and R. K. Jansen. 1995 ndhF sequence evolution and the major clades in the sunflower family. Proceedings of the National Academy of Sciences, USA 92: 10379–10383.
———, ———, R. S. Wallace, H. J. Michaels, and J. D. Palmer. 1992 Phylogenetic implications of rbcL sequence variation in the Asteraceae. Annals of the Missouri Botanical Garden 79: 428–445. [CrossRef][ISI]
Kluge, A. G., and J. S. Farris. 1969 Quantitative phyletics and the evolution of Anurans. Systematic Zoology 18: 1–32.
Kosteletzky, V. F. 1833 Allgemeine Medizinish-Pharmazeutische Flora, vol. 1. Borrosch and André, Prague.
Kroner, G. 1980 Systematische Studien im Umkries von Athrixia Ker Gawl (Asteraceae). Mitteilungen der Botanischen Staatssammlung, München. 16: 1–268.
Linnaeus, C. 1771 Mantissa plantarum altera. Holmiae, Stockholm.
Lundgren, J. 1972 Revision of the genus Anaxeton Gaertn. (Compositae). Opera Botanica 31: 1–59.
———. 1974 The genus Petalacte D. Don (Compositae). Botaniska Notiser 127: 119–224. [ISI]
Maddison, D. R. 1991 The discovery and importance of multiple islands of most parsimonious trees. Systematic Zoology 40: 315–328. [CrossRef]
Maddison, W. P., and D. R. Maddison. 1992 MacClade: analysis of phylogeny and character evolution, version 3.0. Sinauer, Sunderland, Massachusetts, USA.
Margush, T., and F. R. McMorris. 1981 Consensus n-trees. Bulletin of Mathematical Biology 43: 239–244.
Merxmüller, H., P. Leins, and H. Roessler. 1977 Inuleae—systematic review. In V. H. Heywood, J. B. Harbourne, and B. L. Turner [eds.], The biology and chemistry of the Compositae, 577–601. Academic Press, London, UK.
Nordenstam, B. 1989 A synopsis of the genus Syncarpha (Compositae—Gnaphalieae). Compositae Newsletter 17: 2–9.
———. 1996 Anderbergia B. Nord., a new genus of everlastings (Compositae-Gnaphalieae) from the Cape Province. Annalen des Naturhistorischen Museums in Wien 98 (B Suppl.): 403–418.
Oliver, E. G. H., H. P. Linder, and J. P. Rourke. 1983 Geographical distribution of present-day Cape taxa and their phytogeographical significance. Bothalia 14: 427–440.
Puttock, C. F. 1994 Re-analysis of Anderberg's Gnaphalieae data matrix. Compositae Newsletter 25: 1–14.
Rogstad, S. H. 1992 Saturated NaCl-CTAB solution as a means of field preservation of leaves for DNA analyses. Taxon 41: 701–708. [CrossRef][ISI]
Rutherford, M. C., and R. H. Westfall. 1994 Biomes of Southern Africa: an Objective Categorization. Memoirs of the Botanical Survey of South Africa Number 63. National Botanical Institute, Pretoria, South Africa.
Sanger, F., S. Nicklen, and A. R. Coulson. 1977 DNA sequencing with chain-terminating inhibitors. Proceedings of the National Academy of Sciences, USA 74: 5463–5467.
Swofford, D. 1997 PAUP: Phylogenetic analysis using parsimony, pre-release version 4.0.0d64. Laboratory of Molecular Systematics, Smithsonian Institution. Washington, D.C. and Sinauer, Sunderland, Massachusetts, USA.
———. and G. J. Olsen. 1990 Phylogeny reconstruction. In D. M. Hillis and C. Moritz [eds.], Molecular systematics, 411–501. Sinauer, Sunderland, Massachusetts, USA.
Taberlet, P., L. Gielly, G. Pautou and J. Bouvet. 1991 Universal primers for amplification of three non-coding regions of chloroplast DNA. Plant Molecular Biology 17: 1105–1109. [CrossRef][ISI][Medline]
Weimarck, M. J. A. 1941 Phytogeographical groups, centres and intervals within the Cape Flora. Acta Universitatis Lundensis Avd 2. 37, 5: 1–143.
