Plant Syst. Evol. 223:1-13 (2000) Systematics and Evolution © Springer-Vertag 2000 Printed in Austria

Controversial taxonomy of Strumariinae () investigated by nuclear rDNA (ITS) sequences 1. Hessea, Namaquanula, Kamiesbergia, and Dewinterella

Katja Weiehhardt-Kulessa 1, Thomas B6rner 2, Jiirgen Sehmitz 3, Ute Miiller-Doblies 4, and Dietrich Miiller-Doblies 1 ilnstitut ffir Okologie und Biologie, Technische Universitiit Berlin, Berlin, Germany 2Institut fiir Biologie (Genetik), Humboldt-Universit/it zu Berlin, Berlin, Germany SDeutsches Primatenzentrum, G6ttingen, Germany 4Berlin, Germany

Received February 20, 1998 Accepted August 16, 1999

Abstract. Two opposing opinions concerning the phon (subgenus Kamiesbergia) is the sister taxon genetic differentiation of Amaryllidaceae subtribe of H. breviflora. Strumariinae two taxonomic opinions were pub- lished in the last decade. According to Mfiller- Key words: Amaryllidaceae, Strumariinae. Phy- Doblies and Mfiller-Doblies (1985, 1994) the logeny, ITS 1, ITS 2, 18S rDNA, 25S rDNA, Strumariinae includes eight genera, among them maximum parsimony analysis, distance analysis, Hessea, Namaquanula, and Dewinterella. Snijman maximum likelihood analysis. (1991) described the new genus Kamiesbergia and accepted Namaquanula (1992). Of the entire The cosmopolitan family Amaryllidaceae con- complex she recognized only Hessea (Snijman sists of nine tribes, some of them divided 1994). The section Myophila was simultaneously into subtribes (Dahlgren et al. 1985). The pre- described as the genus Dewinterella by Miiller- dominantly South African tribe Doblies and Mfiller-Doblies. is the largest tribe of the family according to The possible phylogenetic relationships of eight the number of genera (11-16 genera) and the species belonging to these taxa are estimated from second in size according to the number of the sequences of the ITS regions of the 18S-25S species. It comprises two subtribes, Amarylli- rDNA. Two species of are used as the dinae (confined to Africa except for the outgroup. The sequences of these taxa are analyzed with maximum parsimony, distance, and maximum subcosmopolitan genus Crinum) and Strumar- likelihood methods. In all phylogenetic reconstruc- iinae (endemic to southern Africa). tions Namaquanula is confirmed to be an inde- Most genera of the Amaryllidinae have pendant clade aside from all other six species of the been revised, either provisionally or compre- Strumariinae. In this group Hessea and Dewinter- hensively, throughout the second and third ella turn out to be sister groups. Hessea stenosi- quarters of this century: Ammocharis and 2 K. Weichhardt-Kulessa: Controversial taxonomy of Strumariinae (Amaryllidaceae)

Cybistetes (Milne-Redhead and Schweickerdt H. bruce-bayeri (D. & U. M-D.) Snijman and 1939), Brunsvigia (Dyer 1950, 1951), Nerine section Myophila with two species, H. pulcher- (Traub 1967), and Crinum in South and East rima (D. & U. M-D.) Snijman and H. mathewsii Africa (Verdoorn 1973, Nordal 1977) and W. E Barker (Fig. 1). When the latter sectional Cameroun (Nordal and WahlstrCm 1980). taxon was published in June 1994, the genus Numerous classifications from the 19th Dewinterella D. & U. M-D. with the same two century dealing with the Hessea-Strumaria species, D. pulcherrima (D. & U. M-D.) D. & U. group and its related genera can be found M-D. and D. mathewsii (W. F. Barker), was (Herbert 1837, Kunth 1850, Salisbury 1866, already in print. As both taxa, the section Baker 1888) which reveal a heterogeneous Myophila and the genus Dewinterella, do not picture of this cluster. Not until the middle of compete for priority, it is possible, therefore, to this century, however, did several botanists consider them as more or less simultaneously start working on it again: Barker (1948), published. Leighton (1948), Obermeyer (1963), Traub The second subgenus, H. subgen. Kamies- (1963), and Goldblatt (1976). In 1985 Miiller- bergia, is monotypic with Hessea stenosiphon Doblies and Mfiller-Doblies validated Traub's (Snijman) D. & U. M-D. (1957) Strumariinae as a subtribe including The subgenus H. subgen. Hessea includes seven genera: Carpolyza Salisb., Namaqua- nine species. Three of them, H. breviflora, nula D. & U. M-D., Strumaria Jacq., Bokke- H. stellaris, and H. cinnabarina, are analyzed veldia D. & U. M-D. (segregated from in the present paper of which the latter species Strumaria), Gemmaria Salisb. (segregated from deserves a taxonomic note. It was not Hessea), Hessea Herb., and Tedingea D. & U. recognized by Snijman and was treated as M-D. Nine years later another genus, Dewinter- synonymous with H. stellaris. The same ella D. & U. M-D., was added (Mfiller-Doblies problem occurred with H. longituba which and Mi.iller-Doblies 1994), accommodating two was placed into the synonymy of H. breviflora. isolated species of Gemmaria. The new genera Thus within the subclade A 1 none of the were accepted by Gunn et al. (1992), Brummitt three additional genera published between (1992), and Greuter et al. (1993). Two of them, 1985 and 1994 are recognized, not even the Tedingea and Namaquanula, were also genus Kamiesbergia of 1991. A single genus, approved by Snijman and Perry (1987) and Hessea, is left (Fig. 1). Snij man (1992). At the beginning of the nineties There are some molecular studies of the a new genus of the Strumariinae, Kamiesbergia Amaryllidaceae and closely related families, Snijman, was published (Snijman 1991) which such as Liliaceae (Shinwari et al. 1994, Fay Miiller-Doblies and Mfiller-Doblies had pub- and Chase 1996). Snijman (1994) presented a lished as a new species of Hessea at the scheme of the phylogenetic relationships of same time. Shortly afterwards Kamiesbergia the Strumariinae based on morphological and stenosiphon was changed to Hessea stenosiphon anatomical data. Because of the controversial (Miiller-Doblies and Mfiller-Doblies 1992). discussion concerning the phylogeny of the Snijman, (1994) who did not recognize Strumariinae, it is useful to look for a new the subtribe Strumariinae of Amaryllideae, source of information for the phylogenetic divided it into two subclades, A 1 (= Hessea reconstruction, based on the results in several sensu Snijman) and subclade A 2 ( --- Carpolyza publications dealing with ITS sequence data as and Strumaria sensu Snijman). The subclade a successful tool to reconstruct the phylogeny A 1 contains three subgenera, Namaquanula, of certain plant groups (e.g. Poaceae by Hsiao Kamiesbergia, and Hessea. Hessea subgen. et al. 1994, 1995a, b; Fabaceae by Wojcie- Namaquanula consists of two sections: section chowski et al. 1993; Asteraceae by Baldwin Namaquanula (formally) with the single species 1992, 1993), this region of the nuclear rDNA, K. Weichhardt-Kulessa: Controversial taxonomy of Strumariinae (Amaryllidaceae) 3

Section Subgenus U Namaq,amda ~ H. bruce.bayeri Nanlaquantda -]__ Section H. pulcherrima Myophila H. mathewsii

Subgenus H. stenosiphon Hessea Kamiesbergia

H. breviflora Subgenus ._~ Hessea Fig. 1. Snijman's dendrogram con- H. stellaris taining the modified subclade A 1 the internal transcribed spacer (ITS) with the of different species and individuals were isolated on intervening 5.8S gene, was chosen. Therefore, different days. it is hoped that this study provides a new and About 1 mg fresh or frozen leaves were homo- conclusive phylogenetic scenario of the Stru- genized in 400gl Wilson-buffer (100raM Tris, mariinae. 10ram EDTA, 100raM NaC1, 0.1% SDS, pH 8.0) with a pestle. The homogenate was treated with RNAse, proteinase K (50 lag/ml), and 1% SDS for Material and methods 2 h at 55 °C. Proteins and cell debris were extracted once with a phenol/chloroform/isoamylalcohol Taxon sampling. Ten species, all arranged in the mix (25 : 24 : 1), once with a chloroform/isoamyl- tribus Amaryllideae of Amaryllidaceae, were used alcohol mix (24: 1) and precipitated in ethanol for the sequence determination of the 5.8S gene (Sambrook et al. 1989). and flanking transcribed spacers (ITS 1 and ITS 2 The investigated DNA region was amplified with regions): Dewinterella pulcherrima, Hessea brevi- primers described by Hsiao et al. (1995a, b). The flora, H. cinnabarina, H. Iongituba, 1t. stellaris, primer ITSL (5 t-TCGTAACAAGGTTTCCG- H. stenosiphon, Namaquanula bruce-bayeri, N. ete- TAGGTG-3') anneals to the 3 ~ end of the 18S sionamibensis, Nerine humilis, and N. sarniensis. rDNA near the ITS 1 border, and the primer ITS4 Both species of Nerine were chosen as the out- (5~-TCCTCCGCTTATTGATATGC-3 ~) is comple- group, because they belong to the Amaryllidinae mentary to the 5 ~ end of the 25S rDNA near the and are therefore closely related to the genera of ITS 2 border. ITSL and ITS4 cover the entire ITS the Strumariinae. Moreover, Nerine was used as region which is divided by the 5.8S gene into the the outgroup in Snijman's work (1994), so that the ITS 1 and ITS 2 regions. results are well comparable. To confirm each sequence, the DNA of two This study was based on living materials of different individuals were sequenced and compared. cultivated from wild provenance. The plants The sequences of each individual were also done in were collected in southern Africa and grown as both directions by using complementary strands. flowers in a greenhouse of the Institute of Ecology The PCR was carried out in a thermal cycler of the Technical University of Berlin. The species, (Biometra) and set for initial 93°C for 3min their abbreviations, taxonomic positions, cultiva- (denaturation), followed by 37 cycles at 93 °C for tion numbers, collectors, collection numbers, grid 30sec (denaturation), 59°C for 30sec (primer numbers, and localities are listed in Table 1. annealing), and 72 °C for 30sec (polymerization). The leaves of two individual plants of each An elongation of the PCR products by 72 °C for species from different locations - when available - 5 min completed the reaction. About 10ng tem- were harvested and stored at -30°C until DNA plate DNA, 400nM primer, 200gM dNTPs, extraction. 1.5 mM MgC12, 10x buffer (100mM TrisHC1, pH DNA extraction and sequencing. To avoid (25°C) 9.0, 500raM KC1, 15mM MgC12, 1% mislabeling or cross-contamination of DNA, DNA's TritonX100, 2 mg/ml BSA or gelatin, 70 °C), and 5 4 K. Weichhardt-Kulessa: Controversial taxonomy of Strumariinae (Amaryllidaceae)

O

O .2° d

©

-.~0

0 ¢~,-~ ~ 00~

£ t~oo ~oo

O

~D O

4- O r,.

O

-.~O 0

O

O 0 5 d 5 ~3 ,.~" o~ d x,_.., .~8 •~

"U. "~ O ~.~ S < K. Weichhardt-Kulessa: Controversial taxonomy of Strumariinae (Amaryllidaceae) 5

U/100 Taq-polymerase (Appligene, Heidelberg) Maximum likelihood trees were constructed by were used. the PHYLIP-DNAML program, version 3.5c The amplified DNA fragment was purified by (Felsenstein 1993). electrophoresis through 1% agarose gel in TAE The comparison of tree length was done by the buffer (0.4M Tris, 0.2M NaAc, 0.01 M EDTA, pH MacClade 3 program (Maddison and Maddison 8.0). QIAquick Gel Extraction Kit (Qiagen, 1992). To be able to compare the tree length of Hilden) was used to purify the gel slice containing Snijman's dendrogram (1994) and the DNA the DNA fragment. The final yield was stored at dendrogram it was necessary to bring both 30 °C. dendrograms into agreement concerning the inves- The purified double-stranded PCR products of tigated species. Therefore only those species all species were directly sequenced on a 373 DNA represented in both studies were included the sequencer (Applied Biosystems, ABI) using Taq calculation of the tree length, so that the number of polymerase and dye-terminators according to the taxa had to be diminished. Seven taxa were then ABI manufacturer's instructions. left: Dewinterella pulcherrima, Hessea breviflora, Sequence alignment and data analysis. The H. stenosiphon, H. stellaris, Namaquanula bruce- sequence data of the roughly 660 bp amplified bayeri, and both species of Nerine. DNA of the investigated species were aligned using CLUSTAL V multiple sequence alignment Results program (Higgins et al. 1992). The parsimony analysis was conducted with the Sequence variation of ITS region. All in all computer program PAUP, version 3.1.1 (Swofford the aligned sequences yielded 662 characters, 1993) using the branch and bound search options to including the 3' end of the 18S gene and the 5' find the most parsimonious tree for the amplified end of the 25S gene. The endpoints of the DNA. All uninformative characters were ignored, regions of the ITS regions and genes were and gaps were treated as missing or as a "fifth based on comparative analysis (Fig. 2). base". Bootstrap values were determined from 500 Table 2 gives information about the total replications. character of the different regions of the The distance analysis of the same aligned amplified DNA and their variable and infor- sequences was conducted with the computer mative positions. program MEGA, version 1.01 (Kumar et al. The length of the complete ITS region of 1993). Gaps were excluded by the pairwise deletion option. Both the Tajima-Nei distance the 10 studied Amaryllidaceae species ranged (Tajima and Nei 1984) and the Kimura 2-parameter from 638 to 640 nucleotides. The ITS 1 region distance (Kimura 1980) were used for the recon- varied from 242 to 245 base pairs (bp) in struction of the MEGA neighbor-joining tree length, and the ITS 2 region spanned from 231 (Saitou and Nei 1987). Bootstraps were done with to 232 bp. The length of the 5.8S gene, 164 bp, 100 replications. was constant in all species (Fig. 2; Table 3).

Table 2. Sequence variation in the ITS region and variable and informative sites of 10 Amaryllidaceae subtribe Strumariinae and Amaryllidinae Region Range from-., to Total character Variable sites Informative sites Total character 1 -.. 662 662 165 124 Total ITS 7... 648 642 164 123 ITS 1 7... 252 246 68 52 5.8S 253...416 164 18 14 ITS 2 417.. • 648 232 78 57 Note: Sites refer to aligned positions in Fig. 2. 6 K. Weichhardt-Kulessa: Controversial taxonomy of Strumariinae (Amaryllidaceae)

Table 3. Sequence length of total character and total ITS and single ITS 1, 5.8S, and ITS 2 regions and G+C percentage of total ITS in 10 Amaryllidaceae subtribe Strumariinae and Amaryllidinae Total Total ITS ITS 1 5.85 ITS 2 character Species length %(G + C) length (in bp) length (in bp) length (in bp) (in bp) D. pulch. 658 638 60 242 164 232 H. brev. • 659 639 56 244 " 231 H. cinn. 660 640 57 .... 232 H. long. 659 639 56 .... 231 H. stel. 660 640 57 .... 232 H. sten. 659 639 56 .... 231 N. b-b. 660 640 55 .... 232 N. etes. " " 55 ...... Ne. hum. " " 60 245 " 231 Ne. sarn. ' ..... 245 .... For species abbreviations see Table 1.

The exact sequence length of the different Dewinterella, D. pulcherrima, resembled the regions and the G+C percentages of the entire sequences of the species of Hessea more than ITS regions for all studied species are given in those of Namaquanula (Fig. 2). Table 3. Hsiao et al. (1994) found the same Phylogenetic analysis. The phylogenetic length of the 5.8S gene in the species of analysis of the investigated species is based on Poaceae and similar G÷C percentages, the alignment shown in Fig. 2. ranging from 57.6% to 64%. Figure 3 presents the most parsimonious Whereas most of the sequence variation tree (PAUP) obtained from the entire ITS occurred in the spacer regions, there was little regions (ITS 1 and 2) with a tree length of 158 variation within the 5.8S subunit. All in all (123 informative positions). The consistency only 18 variable nucleotide positions could be index (CI) value for this tree is 0.854, which is found there (Fig. 2). relatively high and therefore indicated a strong Of the 10 taxa examined, the ITS sequences phylogenetic signal in the data. Gaps were of the two species of Nerine were identical. treated as missing data. Obvious similarities could be found in all The species of Hessea clustered with a sequences of the species of Hessea, of which bootstrapping value of 98%. In this cluster the sequences of Hessea cinnabarina and Hessea cinnabarina and H. stellaris were H. stellaris showed a difference in 11 nucleotide separated based on a bootstrap confidence of positions. The sequences of H. breviflora and 99%. The other three species of Hessea, H. stenosiphon varied in 8, the sequences of H. longituba, H. breviflora, and H. stenosiphon, H. breviflora and H. longituba in 6 nucleotide were grouped with a 100% bootstrap of which positions. There was also considerable H. breviflora and H. stenosiphon were congruence between the two species of Nama- separated from H. longituba by a bootstrap quanula, which differed in 18 nucleotide value of 67%. The separation of Hessea and positions. The sequence of the only species of Dewinterella was supported by a bootstrap ITS i D. pul oh. TCATTGTCGT CGTTTGAATA GATGCTTGCG AACTCGTAGA GCACCTGTAG GC-TCGCAGA 60 If. brev...... CC ...... A ...... T .... GA ...... If. cinn...... CC ...... G ...... GA ...... If. lonq...... C ...... A ...... GA ..... If. stel...... CC ...... G ...... GA ..... If. sten...... CC ...... A ...... GA ..... N. b-b...... GC ...... A .... T..T ...... T...G. .GA ..... N. etes...... G ...... C..A...A .... T..T ...... T...G. .GA...T.• Ne.hum...... CCC ...... A .... T..G ...... GA ...... Ne . sam...... CCC ...... A .... T..G ...... GA ...... 0B0 0 [] DD []

D.pulch. GGCTGTGGCG ATTGCTGCCG CATCCGCCAC CTGGGGTGCC ATTGCCGTTG CCTTCGCCTT 120 If. bre v...... T ...... T...T. .,T ...... If. ci nn...... T. .T ...... If. 1 ong...... T ...... T...T. .T ...... If. stel ...... T. .T ...... If. s ten...... T ...... T...T. .T ...... N. b-b. C.T ...... C ..G.T...T. .C.TT...T. .TA ..... A ..... T.C ...... N. etes. C.T ...... C ..G ..... T. .C..T...T. .TA ..... A ..... T.C ...... Ne. hum...... A ...... C..T...T. .T..T ...... T.C ...... Ne. sam...... A ...... C..T...T. .T..T ...... T.C ...... DD OD [] [3 [3 [] [] ~D D 0B

D.pulch. GCA-TGGCTG CGGGAGAGGG -TAGTGGGAA CAA-CATCCG GCGCGTCGTG CGCCAAGGAG 180 If.brev...... C ...... A .... T. .T..AGT ..... T ...... H. cinn...... T ...... A ...... AGTA .... T ...... If. long...... C ...... A .... T ..... AGT ..... T ...... If. stel...... T.. T ...... A ...... AGTA .... T ...... If. sten...... C ...... A .... T ..... AGT ..... T ...... N.b-b. ...CC..TA. .A .... G .... A ...... ATG..T. .T...GT ...... N. etes. ...CC..TA. .A .... G .... A ...... ATG ..... T...GT ...... Ne.hum. ...C ...... A .... G... GC ...... A ...... GT ...... T. Ne.sarn. ...C ...... A .... G... GC ...... A ...... GT ...... T. [] N[3 O D [3 13 ~D 13 F? [] [] []

D. pul ch. CAAGACCTGT TGGAGAGCAG AGCGTGCTGG CATGCTAGTT GCTCGAGCTT GCGATGCGAT 240 H.brev. .G ...... T ...... G. ...T ...... T... If. cinn. • .T ...... A ..... T .... A ...... G. •..T ...... A IT. long. .G ...... T ...... G. • . .T ...... T... If. stel. • .T..T ...... A ..... T .... A ...... G. • . .T ...... i H. sten. .G ...... T ...... G. • . .T ...... T... N. b-b. .T ..... A ...... C.A .G ...... GC N. etes. .T ..... A ...... C.. .G ...... G. .T.G ...... Ne.hum...... T .... C.. .G ...... G. • . .T ...... C ..... ATe. sarn...... T .... C.. .G ...... G. •..T ...... C ..... [3G [] U [] [3 [] [][3 [3[3 []

6 5.8S D.pulch. CTTTGGTACT TCATTACGAC TCTCGGCAAC GGATATCTTG GCTCTCGCAT CGATG~GGA 300 If. bre v. .... C ...... i ...... If. ci nn...... A ...... T ...... If. long• .... C ...... i ...... IT. stel...... A ...... IT. s ten. .... C ...... A.A ...... N.b-b. . . .- ...... G.A.T ...... T ...... C ...... TGA ...... N.etes. • .•-...G .... G.G.T ...... T ...... C ...... TGA ...... ATe. hum. ...- ...... i ...... Ne. sam. • ..- ...... i ......

[] D D D 0 UBD

D.pulch. CGTAGCG~A TGCGATACTT GGTGTG~TT GCAG~TCCC GTG~CCATC GAGTCTTTGA 360 H. brev. H. cinn. H. long. H. stel. ::::::::U :::::::::: 1111121111 :::::::::: 1111211111 1111211111 H. sten. N.b-b. 12221111;i 2122222212 1111112121 2221211122 2222222222 ::::~l:::: N.etes...... G ...... T ..... Ne.hum. Ne.sarn.

S Fig. 2. Aligned sequences of 10 taxa of Amaryllidaceae subtribe Strumariinae and Amaryllidinae Numbers indicate the succeeding positions of 1 to 662 from the end of the 18S region to the beginning of the 25S region; arrows mark the beginning of ITS 1, 58S, and ITS 2 regions; dashes denote gaps; dots indicate identity to Dewinterella pulcherrima; blank rectangles mark informative sites 8 K. Weichhardt-Kulessa: Controversial taxonomy of Strumariinae (Amaryllidaceae)

ITS 2 D.pulch. ACGCAAGTTG CGTCCGAGGC TATCTGGCTA AGGGCACGCC TGCCTGGGCA TCACGCCTCG 420 H, brev...... C ...... C ...... T ...... H. cinn...... C ...... H. long...... C ...... C ...... T ...... H. stel...... C ...... H. sten...... C.T ..... C ...... N.b-b...... C.T ...... T ...... T ...... T .... N. etes...... C.T ...... T ...... T ...... T .... Ne.hum...... C ...... Ne. sarn...... C ...... 0 [] [] [] 0

D. pul oh. TGACGCTTCG TGCCATCTGC CCCCCACCTG GTGCTGGTGA CAACTGGCGC GAACGCGGGG 480 If. brev...... T ..... C..A .... T ...... A .... G..T .... A ...... T .... I{. cinn...... T ..... C..A .... T ...... A .... G ...... If. 1 ong...... T ..... C..A .... T ...... A .... G..T...T.T ..... T .... If. s tel...... T. ..T.C..A .... T...T ...... A .... G ...... If. sten. @ ...... T ..... C..A .... T ...... A .... G..T .... A ...... T .... N. b-b...... T ..... C ..... GTTT .... T ...TA..C.. T ...... T .... T..T.CA. N. etes...... T ..... CT..T .GTTT .... T .... A.,C.. T ...... T .... T..T.CA. Ne. hum...... C ...... T ...... A. .C ...... T ...... TC.A. Ne. sam...... C ...... T ...... A. .C ...... T ...... TC.A. 0 0 O0 0 0 O0 0 0 0 O0 [] 0000

D. pul ch . ACTGGCCCTC TGTGCCTCGT CGTGCGGT@G GTTAAAGTGT GCGTCGTTGG CGGGTCGGAT 540 II.brev. .T ...... G.CT ...... C.T.. }I. cinn...... C ...... G.CT ..... T...C.T.. H. lonq. .T ...... G.CT ...... C.T.. }I. stel ...... C ...... G.C ...... C.T.. H. s ten. .T ...... G.CT ...... CTT.. N. b-b. .T ...... CT ...A.G..A ...... A ...... G ...... T...CT... N. etes. .T ...... CT , . .A.G..A ...... A ...... G ...... T,..CT... Ne.hum. -T ...... C ...... G..A ...... @,C...C ...... C .... Ne. sarn. -T ...... C ...... G..A ...... G.C...C ...... C .... [] O0 0 0 0 [] O0 0 0 O0

D.pulch. GCGGCGAGTG GTGGAGAACA CACGCACGAC GTCGTTGGAG ATGCCCAGCC CAGAACGGTG 600 IT. brev...... T ...... T ...... A ...... T.C..T .T ...... H. cinn...... T ...... T ...... A ...... T.T ..... T ...... I{. long...... T ...... T ...... A ...... T.C..T .T ...... If. stel...... T ...... T..A ..... A ...... T.T ..... T ...... If. sten. .... TT ...... T ...... A ...... T.C..T .T ...... N. b-b. .... T ...... A...T..T ...... T ..... TA.. AT...T .... N. etes. .... T ...... T.A ...... T ...... T ..... TA.. AT...T .... Ne.hum...... C .... C... T ..... T... TT ..... T.. Ne . sam...... C .... C... T ..... T... TT ..... T.. OO O 0 0 [] 0 0 [] 0 O0 0 000 0 []

~, 25S D.pulch. CGTTGGAGGG ATCCACGTGG GTGGGCGCAA GTTGAGCGCC CTTAGAACAA GATCCCAGGT 660 H.brev. .A..A. ..A ...... CT. ..C ...... T..G..T-...... E. cinn. .A..A. ..A ...... CT. . .C ...... G ...... ~.lonG. .A..A. . .A ...... CT. ..C ...... T..G..T-...... H. stel. .A..A. ..A ..... G.CT. • .C ...... T..G ...... ~.sten. .A,.A. ..A ...... CT. ..C ...... T..G..T-...... N.b-b. .A ...... A ...... T...... G ..... AT .... G..G ...... N. etes. .A ...... A ...... T...... T ..... GG .... AT .... G..G ...... Ne.hum. .A ...... A ...... CT, ...... G ..... T ...... Ne.sarn. .A ...... A ...... CT...... G ..... T ...... 0 [] 0 00 [] 0 0

D. pu l ch . CA II.brev. If, cinn. H. long. H. stel. I{. sten. N. b-b. N. etes. Ne.hum. Ne . sam.

Fig. 2 (continued) K. Weichhardt-Kulessa: Controversial taxonomy of Strumariinae (Amaryllidaceae) 9

D.pulch. bootstrap confidence of 100%. The separation of H. breviflora and H. stenosiphon from l l~H.brev. H. longituba was sustained by a bootstrap 23 value of 74% or 82%. Hessea and Dewinterella ( H.sten. were separated, supported by a 100% bootstrap. The same bootstrap value indicated the mono- 17 3 2 H.long. phyly of Namaquanula which turned out to be a sister group of Hessea/Dewinterella with a 100% bootstrap confidence. The PHYLIP maximum likelihood tree 2 . . (not shown) further confirmed the topology of the parsimony tree. 2 N.b-b. 55 The most parsimonious trees (PAUP) obtained only from the single ITS 1 and ITS 2 ~ N.etes. 18 I regions both had a high CI value (0.894 and 0.813) which pointed to a strong phylogenetic Ne.hum. signal in the data. Both trees turned out to be Ne.sarn. identical in their tree topology compared to the other trees. Only in some species of Fig. 3. Most parsimonious tree obtained from the Hessea did both trees present polytomies. entire ITS region. Numbers in circles denote Tree length analysis. The comparison of bootstrap percentages; numbers above the branches the tree length between the two different indicate the inferred branch-length excluding uninformative characters dendrograms (Figs. 1, 3) showed that 28 additional steps would be necessary to impose Snijman's tree topology on the molecular data. confidence of 100%. The monophyly of Discussion Namaquanula was confirmed by a 100% boot- strap confidence. The separation of Hessea/ According to Snijman's subclade A 1 Hessea Dewinterella and Namaquanula from a com- mathewsii and H. pulcherrima (= Dewinterella mon ancestor was sustained by a 100% pulcherrima) are presented as a sister group of bootstrap. H. bruce-bayeri (= Namaquanula bruce - bay- The same tree topology gained from the eri) (Fig. 1). The DNA phylogeny, however, whole ITS region occurred when gaps were showed that Dewinterella is a sister group of treated as a "fifth base". Hessea, indicated by a 100% bootstrap con- The MEGA neighbor-joining trees based fidence. Namaquanula turns out to be the sister on the analysis of the entire ITS region revealed group to Dewinterella and Hessea which is also exactly the same tree topology compared to the sustained by the maximum bootstrap value of tree obtained from the parsimony analysis. 100% (Fig. 3). The result of this parsimony Both trees (not shown), the one done with the analysis was confirmed by the same results Tajima-Nei distance and the other one done gained from the distance analysis and max- with the Kimura-2-parameter distance, showed imum likelihood trees. that the species of Hessea were grouped with a This result indicates that the generic rank bootstrapping value of 98% or 99%. H. stellaris should be used instead of the subgeneric and and H. cinnabarina on the one hand and sectional rank for all three taxa. H. breviflora, H. stenosiphon, and H. longituba Figure 4 compares of Snijman's modified on the other hand, each clustered with a dendrogram (Fig. 1) on the left hand side and 10 K. Weichhardt-Kulessa: Controversial taxonomy of Strumariinae (Amaryllidaceae)

F N.b-b. Section [-" - - Namaquanula - 1 I" Naokaquanula : t I Subgenus I !,.. N.etes. --I i r-- Nan~aquanula-I I I t- Section i I Myophila II , D.pulch. - - - - Dewinterella - 1 1._..I ! I I ! I ! ! I I I I I Hessea~Kamiesbergia! Subgenus H.sten. jA i J

H.brev.

L,, Subgenus H.Iong. ~ Hessea Hessea H.cima. --] H.stel. Fig. 4. Comparison of Snijman's dendrogram on the left hand side and the dendrogram based on the DNA phylogeny on the right hand side. Bold lines mark the species investigated on both sides; bold lines also unite the taxa Hessea subgen. Hessea and subgen. Kamiesbergia on the left and the taxon Hessea s.str. (sensu Miiller-Doblies and MiJller-Doblies) on the right side; broken line shows H. subgen. NamaquanuIa on the left and Namaquanula and Dewinterella on the right side; dotted line indicates that this species is represented in Snijman's work but not described by her; asterisks indicate species which are recognized by Miiller-Doblies and Mtiller-Doblies but not by Snijman the dendrogram based on the DNA phylogeny In this context the question has to be raised (Fig. 3) on the right hand side. Different types which of the two patterns uniting the three of lines are used which link taxa in Snijman's genera, Hessea, Dewinterella, and Namaqua- dendrogram and correspond to the taxa in the nula, is to be preferred, the dendrogram of the DNA dendrogram so that another fact DNA analysis or the one based on Snijman's becomes obvious. Either the monotypic sub- suggestion (Figs. 3, 1). To begin with, the genus Kamiesbergia has to be sunk into strong signals in the DNA (e.g. high bootstrap Hessea s.str, because the DNA analysis proved values, the high number of informative sites, Hessea stenosiphon to be the sister taxon of H. high consistency index) strongly support the breviflora, or the genus Hessea needs to be reconstructed phylogeny. Furthermore, the subdivided into at least four subgenera just to comparison of the tree lengths of both accommodate the five investigated species of dendrograms on the basis of the sequences Hessea. proved the DNA dendrogram to be shorter and Despite the conviction that Hessea brevi- therefore more probable. If one also takes the flora and H. stellaris are sister taxa (Snijman vegetative characters, bulb structures, and 1994; Fig. 1), the DNA phylogeny does not floral characters of these three genera into support this proposal, for these two species do consideration, the affinities are very clear. not cluster as a sister group. All molecular Hessea and Dewinterella have more character- analyses showed that there are two groups of istics in common, e.g. two foliage leaves and a the studied species of Hessea, H. breviflora, sheathing cataphyll, than Dewinterella and H. stenosiphon, and H. longituba on the one Namaquanula, whose foliage leaves differ hand and H. cinnabarina and H. stellaris on in number from one to five and which lack the other hand (Fig. 3). a sheathing cataphyll (Table 4). The only K. Weichhardt-Kulessa: Controversial taxonomy of Strumariinae (Amaryllidaceae) 11

Table 4. Morphological comparison between three genera Hessea, Dewinterella, and Namaquanula Hessea and Dewinterella Namaquanula Vegetative characters Foliage leaves 2 1-5 Leaf in cross section canaliculate elliptic Bulb structure Prophyll usually present absent Sheathing cataphyll present absent Floral characters Perigon stellate (when fully expanded) funnel-shaped Perigon tube absent or pterotube eutube

common characters between Namaquanula Brummitt R. K. (1992) families and and Dewinterella are the filament appendages genera. Royal Botanic Gardens, Kew. and the adjacent papillae at the filament bases Dahlgren R. M. T., Clifford H. T., Yeo E E (1985) (Mfiller-Doblies and Mfiller-Doblies 1985, The families of the monocotyledones: structure, 1994). This morphological comparison stress- evolution, and taxonomy. Springer, Berlin Hei- delberg New York Tokyo. es the result obtained from the DNA phylo- Dyer R. A. (1950) A review of the genus geny, namely that there is a closer relation Brunsvigia. P1. Life 6: 63-83. between Hessea and Dewinterella than Dyer R. A. (1951) A review of the genus between Dewinterella and Namaquanula. Brunsvigia Heist. P1. Life 7: 44-64. Fay M. E, Chase M. W. (1996) Resurrection of We thank Prof. Dr. Robin E A. Moritz for Themidaceae for the Brodiaea alliance, and providing the laboratory space and also acknowl- recircumscription of Alliaceae, Amaryllidaceae edge Dr. Andreas Weihe's profound introduction and . Taxon 45: 441-451. into the computer analysis. Felsenstein J. (1993) PHYLIP (Phylogeny infer- ence package), version 3.5c. Seattle. Goldblatt E (1976) Chromosome cytology of References Hessea, Strumaria, and Carpolyza (Amaryllida- ceae). Ann. Missouri Bot. Gard. 63: 314- Baker J. G. (1888) Handbook of the Amaryllideae, 320. including the Alstroemerieae and Agaveae. Greuter W., Brummitt R. K., Farr E., Kilian N., George Bell, London. Kirk E M., Silva E C. (1993) Names in current Baldwin B. G. (1992) Phylogenetic utility of the use for extant plant genera. Koeltz Scientific internal transcribed spacers of nuclear ribosomal Books, K6nigstein. DNA in plants: an example from the Composi- Gunn C. R., Wiersema, J. H., Ritchie C. A., tae. Molec. Phylogenet. Evol. 1: 3-16. Kirkbride J. H. Jr. (1992) Families and genera of Baldwin B. G. (1993) Molecular phylogenetics of Spermatophytes recognized by the Agricultural Calycadenia (Compositae) based on ITS Research Service. USDA, Beltsville, Maryland. sequences of nuclear ribosomal DNA: chromo- Herbert W. (1837) Amaryllidaceae. James Ridg- somal and morphological evolution reexamined. way, London. Amer. J. Bot. 80: 222-238. Higgins D. G., Bleasby A. J., Fuchs R. (1992) Barker W. E (1948) Plantae novae africanae XX. CLUSTAL V: improved software for multiple J. S. African Bot. 14: 29-40. sequence alignment. Cabios 8: 189-191. 12 K. Weichhardt-Kulessa: Controversial taxonomy of Strumariinae (Amaryllidaceae)

Hsiao C., Chatterton N. J., Asay K. H., Jensen K. B. Nordal I., Wahlstrc~m R. (1980) A study of the (1994) Phylogenetic relationships of 10 grass genus Crinum (Amaryllidaceae) in Cameroun. species: an assessment of phylogenetic utility of Adansonia, ser. 2 20: 179-198. the internal transcribed spacer region in nuclear Obermeyer A. A. (1963) Hessea tenella. F1. P1. ribosomal DNA in monocots. Genome 37: Africa 36 ("1964"): t. 1413. 112-120. Saitou N., Nei M. (1987) The neighbor-joining Hsiao C. (1995a) Molecular phylogeny of the method: a new method for reconstructing Pooideae (Poaceae) based on nuclear rDNA phylogenetic trees. Molec. Biol. Evol. 4: (ITS) sequences. Theor. Appl. Genet. 90: 406-425. 389-398. Salisbury R. A (1866) The genera of plants: a Hsiao C. (1995b) Phylogenetic relationships of the fragment containing part of Liriogamae. John monogenomic species of the wheat tribe, van Voorst, London. Triticeae (Poaceae), inferred from nuclear rDNA Sambrook J., Fritsch E. F., Maniatis T. (1989) (internal transcribed spacer) sequences. Genome Molecular cloning, Cold Spring Harbor 38: 211-223. Laboratory, New York. Kimura M. (1980) A simple method for estimating Shinwary K. Z., Terauchi R., Utech F. H., Kawano evolutionary rate of base substitutions through S. (1994) Recognition of the New World comparative studies of nucleotide sequences. Disporum section Prosartes as Prosartes (Lilia- J. Molec. Evol. 16: 111-120. ceae) based on the sequence data of the rbcL Kumar S., Tamura K., Nei M. (1993) MEGA: gene. Taxon 43: 353-366. Molecular evolutionary genetics analysis, ver- Snijman D. A. (1991) Kamiesbergia, a new sion 1.01. Pennsylvania. monotypic genus of the Amaryllideae - Strumar- Kunth C. S. (1850) Enumeratio plantarum omnium iinae (Amaryllidaceae) from the north-western hucusque cognitarum, secundum familias natur- Cape. Bothalia 21: 125-128. ales disposita, adjectis characteribus, differentiis Snijman D. A. (1992) Review of the systematics of et synonymis 5: Enumeratio Asparaginearum. the Strumariinae (Amaryllidaceae). Herbertia J. C. Collae, Stuttgart, Tiibingen. 48: 52-57. Leighton F. M. (1948) Some changes in nomencla- Snijman D. A. (1994) Systematics of Hessea, ture:V. J. S. African Bot. 14: 81-83. Strumaria and Carpolyza (Amaryllideae: Ama- Maddison W. R, Maddison D. R. (1992) Mac- ryllidaceae). Cape Town, Bolus Herbarium. Clade: Interactive analysis of phylogeny and Snijman D. A., Perry R (1987) A floristic analysis character evolution. Sunderland. of the Nieuwoudtville Wild Flower Reserve, Milne-Redhead E., Schweickerdt H. G. (1939) A north-western Cape. S. African J. Bot. 53: new conception of the genus Ammocharis Herb. 445-454. J. Linn. Soc., Bot. 52: 159-197. Swofford D. L. (1993) PAUP: phylogenetic analysis Miiller-Doblies D., Miiller-Doblies U. (1985) De using parsimony, version 3.1.1. Champaign. Liliifloris notulae: 2. De taxonomia subtribus Tajima E, Nei M. (1984) Estimation of evolu- Strumariinae (Amaryllidaceae). Bot. Jahrb. Syst. tionary distance between nucleotide sequences. 107: 17-47. Molec. Biol. Evol. 1: 269-285. Miiller-Doblies D., Miiller-Doblies U. (1992) De Traub H. R (1957) Classification of the Amarylli- Liliifloris notulae: 4. Notes on Hessea (including daceae: subfamilies, tribes and genera. P1. Life Kamiesbergia; Amaryllidaceae of southern 13: 76-83. Africa). Gleditschia 20: 15-20. Traub H. R (1963) The genera of Amaryllidaceae. Miiller-Doblies D., Miiller-Doblies U. (1994) De The American Plant Life Society, La Jolla, Calif, Liliifloris notulae: 5. Some new taxa and Traub H. R (1967) Review of the genus Nerine. P1. combinations in the Amaryllidaceae tribe Ama- Life 23:1-32 (Suppl.). ryllideae from arid southern Africa. Feddes Verdoorn I. C. (1973) The genus Crinum in Repert. 105: 331-363. southern Africa. Bothalia 11: 27-52. Nordal I. (1977) Revision of the East African taxa Wojciechowski M. E, Sanderson M. J., Baldwin B. of the genus Crinum (Amaryllidaceae). Norweg. G., Donoghue M. J. (1993) Monophyly of J. Bot. 24: 179-194. aneuploid Astragalus (Fabaceae): evidence from K. Weichhardt-Kulessa: Controversial taxonomy of Strumariinae (Amaryllidaceae) 13

nuclear ribosomal DNA internal transcribed Germany. Jiirgen Schmitz, Deutsches Primatenzen- spacer sequences. Amer. J. Bot. 80: 711-722. trum, Kellnerweg 4, D-37077 G6ttingen, Germany. Thomas B6rner, Institut ffir Biologie (Genetik), Addresses of the authors: Katja Weichhardt- Humboldt-Universitfit zu Berlin, Chausseestrasse Kulessa, Dietrich Mfiller-Doblies, Institut f/Jr 117, D-10115 Berlin, Germany. Ute Miiller- Okologie und Biologie, Technische Universit~it Doblies, Paulsenstrasse 47a, D-12163 Berlin, Berlin, Franklinstrasse 28/29, D-10587 Berlin, Germany.