Plant Syst. Evol. 228: 33±48 22001)

A reassessment of Normania and Triguera )

L. Bohs1 and R. G. Olmstead2

1Department of Biology, University of Utah, Salt Lake City, UT, USA 2Department of Botany, University of Washington, Seattle, WA, USA

Received August 8, 2000 Accepted April 2, 2001

Abstract. Normania and Triguera comprise two investigation of speciation, adaptive radiation, genera of the Solanaceae whose anities have been morphological specialization, and long dis- uncertain. Normania encompasses two tance dispersal in . Island endemics often endemicto Macaronesia; Triguera is monotypic exhibit highly divergent morphologies com- and found in Spain and northwestern Africa. Both pared to their mainland relatives and in many have slightly zygomorphic¯owers and horned cases these disjunct but evolutionarily close anthers that dehisce by both apical pores and relationships have been clari®ed only with the longitudinal slits. Micromorphological similarities include trichotomously colporate pollen grains and recent advent of molecular data. Such is the seed surface cells with radially thickened exten- case with several endemic Macaronesian taxa sions. Molecular data from the chloroplast ndhF of the Solanaceae. Olmstead and Palmer gene and the nuclear ITS region establish that 21997) included the Macaronesian endemic Normania and Triguera are nested within the large vespertilio in their phylogeneticstudy genus Solanum, where together they form a well based on chloroplast DNA restriction frag- supported clade. However, the relationship of this ment variation, but the systematicplacement clade to other Solanum subgroups is not resolved. of the other endemicsolanaceoustaxa has not Transfer of the Normania and Triguera epithets to been examined using molecular data and their Solanum is made, necessitating one new name. The nearest relatives have not been identi®ed with molecular data con®rm that the species of Solanum certainty. endemic to Macaronesia belong to two distinct The purpose of this study is to elucidate the clades, each showing an independent evolution of heteromorphicanthers. systematicposition of two small and enigmatic genera of the Solanaceae, Triguera and Nor- Key words: Normania, Triguera, Solanum, Macaro- mania. Triguera is monotypicand native to the nesia, ndhF, ITS. Iberian peninsula and northwestern Africa. Normania includes two species endemic to the Since the time of Darwin 21845), oceanic Macaronesian islands of Madeira and the islands have served as living laboratories for Canaries. Both have been placed in subfamily the study of evolutionary questions. Research- Solanoideae, tribe Solaneae, which includes ers have targeted archipelagos such as the genera with ¯attened seeds, curved embryos, Galapagos, Macaronesia, and the Hawaiian generally valvate corolla aestivation, and and Juan Fernandez Islands as sites for basi®xed anthers. Because of their unusual 34 L. Bohs and R. G. Olmstead: A reassessment of Normania and Triguera 2Solanaceae) distribution and aberrant morphology, the apparently elongate into longitudinal slits. The relationships of these two genera have been style is straight, ca. 5±7 mm long and included, obscure and misunderstood. Molecular data and the stigma is minute. The are now o€er the opportunity to examine the globose berries with a dry or membranous phylogeneticposition of these taxa within the texture, ca. 10 mm in diameter, and are Solanaceae and clarify the close anity be- subtended by the somewhat accrescent folia- tween them, as has been suggested by previous ceous calyx. Each contains 4±6 large, authors 2Lowe 1872, Francisco-Ortega et al. dark brown, deeply pitted seeds. 1993). All previous Solanaceae taxonomists have Triguera osbeckii was ®rst described by recognized Triguera as a distinct genus. Miers Linnaeus 21753) as a species of Verbascum,a 21849a) suggested that Triguera is closely allied genus later assigned to the Scrophulariaceae. with Solanum on the basis of its stamen In 1786, Cavanilles erected the genus Triguera structure and corolla aestivation. He reiterates and described the new species T. ambrosiaca this view in another paper 2Miers 1849b), in Cav. and T. inodora Cav. The genus Triguera which he lists Triguera in the tribe Solaneae Cav. 21786; Solanaceae) is conserved over along with the genera Solanum, Lycopersicon, Triguera Cav. 21785; Bombacaceae; Farr et al. and . D'Arcy 21991) likewise 1979). Gmelin in 1791 described another new associated Triguera with other poricidally species in the genus, T. baccata J. F. Gmel. dehiscent genera such as Solanum, Lycopersi- Subsequent authors 2e.g. Poiret 1808, Roemer con, Cyphomandra, and Lycianthes. Molecular and Schultes 1819, Sprengel 1825, Miers phylogeneticstudies have resulted in the 1849a, Willkomm 1870, Hawkes 1972) recog- placement of Lycopersicon and Cyphomandra nized two to three species in Triguera, but within Solanum, and have established that Hansen and Hansen 21973) consider it likely Lycianthes is probably distinct from Solanum that the genus is monotypic, with T. osbeckii 2Spooner et al. 1993; Bohs 1995; Bohs and 2L.) Willk. as the only species. However, Olmstead 1997, 1999; Olmstead and Palmer Hansen and Hansen 21973) were unable to 1992, 1997; Olmstead et al. 1999). None of reach a conclusion about the identity of these molecular studies have examined the T. inodora, whose type has not been located. phylogeneticplacementof Triguera. In all probability, T. inodora is either a The unusual distribution and ¯oral mor- synonym of T. osbeckii or represents a taxon phology of Normania have attracted the unrelated to Triguera. attention of several previous workers. The Triguera osbeckii ranges from southern genus consists of two species, N. nava and Spain to adjacent northern Africa in Morocco N. triphylla, both endemicto Macaronesia. and northwestern Algeria. It is a small herb or Normania nava is one of the rarest species of weakly woody , apparently annual, with Macaronesia and is restricted to the islands of alternate, sessile, obovate, coarsely dentate Tenerife and Gran Canaria in the Canary leaves. The ¯owers are solitary or paired in Island archipelago. Only two living plants of the leaf axils, and have rather large, foliaceous this species have been found since the time of calyces densely covered with curled, un- its original description in the ®rst half of the branched, whitish hairs. The rotate-campanu- nineteenth century 2Francisco-Ortega et al. late corolla is dark purple, shallowly ®ve-lobed 1993). Normania triphylla is likewise rare in at the apex, and slightly zygomorphic. The ®ve its natural range on the island of Madeira. stamens are equal or subequal in size and However, seeds gathered from a single wild shape, the ®laments short 2ca. 1 mm long), in 1994 were taken to the National with short 24±5 mm) anthers that dehisce by Botanical Conservatory in Brest, France, terminal pores located beneath two small where plants were successfully cultivated apical horns. As the anthers age, the pores and have even became locally naturalized L. Bohs and R. G. Olmstead: A reassessment of Normania and Triguera 2Solanaceae) 35

2R. Lester, pers. comm.). Seeds have been Dunal because of their supposed basal pedicel distributed to various botanicgardens and articulation. Lowe 21872) later transferred Solanaceae specialists. these species to his new genus Normania and Like Triguera osbeckii, the species of proposed that they were closely related to the Normania are herbs to weakly woody short- genus Triguera. Later botanists did not recog- lived . The stems and leaves are covered nize Normania and included N. triphylla and with soft, unbranched, often glandular hairs. N. nava within Solanum 2Bentham 1876, Bitter The leaves of both Normania species are 1912, D'Arcy 1972, Child 1990). Bitter 21912) membranaceous and elliptic-ovate in outline, disagreed with placement of the two species with those of N. triphylla usually pinnately near the potato group 2section Tuberarium lobed or dissected. The in¯orescences are Bitter) and moved them into his new section pedunculate, unbranched, and relatively few- Normania. D'Arcy 21972) and Child 21990) 2less than 15-) ¯owered. Calyces are large, retained Bitter's section Normania, but con- leafy, and soft-pubescent as in Triguera sidered it to be included in subgenus Potatoe osbeckii. The corolla is purple, rotate-campa- 2G. Don) D'Arcy. Most recently, Francisco- nulate, shallowly ®ve-lobed, and slightly zygo- Ortega et al. 21993) have supported recogni- morphic2Fig. 1). Despite these similarities tion of Normania as a separate genus within with Triguera, the ¯owers of both Normania the tribe Solaneae due to its distinctness in species are distinct due to their remarkable macro- and micromorphological characters. stamens. The ®ve anthers di€er greatly in size They suggested a close relationship with and structure: two are long 26±11 mm) and Triguera on the basis of similarities in overall curved, two are shorter 24.5±8.5 mm) and also morphology and in pollen and seed structure. curved, and one is quite short 23±4.5 mm). The Whether recognized as Normania or consid- four longer anthers have a projection or horn ered as Solanum, the phylogeneticposition of at the middle or near the base. Although a N. triphylla and N. nava has not been resolved. small pore is apparent near the tips of the Four of the solanaceous species endemic to longer anthers, they mainly dehisce by a Macaronesia 2Normania nava, N. triphylla, longitudinal slit that develops from near the Solanum vespertilio, and S. liddii Sunding) base of the pore and opens proximally. have anthers that are markedly unequal in Normania nava and N. triphylla have minor size. Two of these are included in the di€erences in the color and form of their molecular analyses reported here. Triguera anthers 2R. N. Lester, pers. comm.), but both has equal or subequal anthers, but has been are similar in the overall morphology of the suggested as a possible relative of Normania by androecium. Unlike Triguera, the style in Lowe 21872) and Francisco-Ortega et al. Normania is long 2ca. 11±12 mm) and curved 21993). This study thus a€ords the opportunity and extends through the two longest anthers. to ascertain whether heterandry 2unequal The fruits are bright red to orange, globose, anthers within a single ¯ower) evolved con- somewhat ¯eshy, and subtended by the accres- vergently in separate lineages, as suggested by cent calyx 2Fig. 2). Plants cultivated at the Lester et al. 21999). University of Utah greenhouse were self- compatible and autogamous. Normania triphylla was originally placed in Materials and methods the genus Nycterium Vent. by Lowe 21838). The data presented here are a subset of a larger Webb and Berthelot 21845) originally de- analysis of over 100 species of Solanum and scribed Normania nava as a species of Solanum. related genera. In this paper, we present data Dunal 21852) considered them both to belong from 40 species of Solanaceae, including taxa to Solanum, and placed them in section representing broad sampling among subgroups of Pachystemonum Dunal subsection Tuberarium Solanum and selected outgroups from subfamily 36 L. Bohs and R. G. Olmstead: A reassessment of Normania and Triguera 2Solanaceae)

Solanoideae. Outgroup taxa were chosen based on Provenance and voucher information is given in previously published phylogeneticanalyses of Table 1. Olmstead and Palmer 21992, 1997), Bohs and DNA was extracted from fresh or silica dried Olmstead 21997), and Olmstead et al. 21999). leaf samples using the modi®ed CTAB method of

Figs. 1±2. Fig. 1. of Solanum trisectum.Scalebarˆ 1cm.Fig. 2. Fruits of Solanum trisectum.Scale bar ˆ 1cm L. Bohs and R. G. Olmstead: A reassessment of Normania and Triguera 2Solanaceae) 37

Table 1. Sources of taxa sequenced for ndhF and ITS Taxon Sourcea Voucherb GenBank accession numbers ndhF ITS

Capsicum baccatum L. var. pendulum 2 Eshbaugh 1584 U08916 AF244708 2Willd.) Eshbaugh Jaltomata procumbens 2Cav.) J. L. Gentry 3 Davis 1189A U47429 AF244710 Lycianthes heteroclita 2Sendtn.) Bitter 1 Bohs 2376 U72756 AF244709 Normania triphylla 2Lowe) Lowe 1 Bohs 2718 AF224063 AF244733 Physalis alkekengi L. 2 D'Arcy 17707 U08927 AF244711 Solanum abutiloides 2Griseb.) Bitter & Lillo 2 RGO S-73 U47415 AF244716 Solanum adhaerens Roem. & Schult. 1 Bohs 2473 AF224061 AF244723 Solanum allophyllum 2Miers) Standl. 1 Bohs 2339 U47416 AF244732 Solanum appendiculatum Dunal 2 Anderson 1401 AF224062 AF244746 2CONN) Solanum arboreum Dunal 1 Bohs 2521 U47417 AF244719 Solanum argentinum Bitter & Lillo 1 Bohs 2539 U72752 AF244718 Solanum aviculare G. Forst. 2 BIRM S.0809 U47418 AF244743 Solanum betaceum Cav. 1 Bohs 2468 U47428 AF244713 Solanum campechiense L. 1 Bohs 2536 AF224071 AF244728 Solanum candidum Lindl. 2 RGO S-100 AF224072 AF244722 Solanum cordovense Sesse & MocË .1Bohs 2693 U72751 AF244717 Solanum dulcamara L. 2 none U47419 AF244742 Solanum elaeagnifolium Cav. 2 RGO S-82 AF224067 AF244730 Solanum glaucophyllum Desf. 2 none U72753 AF244714 Solanum jamaicense Mill. 2 RGO S-85 AF224073 AF244724 Solanum laciniatum Aiton 1 Bohs 2528 U47420 AF244744 Solanum luteoalbum Pers. 1 Bohs 2337 U72749 AF244715 Solanum lycopersicum L. 2 none U08921 AF244747 Solanum macrocarpon L. 2 RGO S-88 AF224068 AF244725 Solanum mammosum L. 2 RGO S-89 AF224074 AF244721 Solanum melongena L. 2 RGO S-91 AF224069 AF244726 Solanum nitidum Ruiz & Pav. 1 Nee 31944 2NY) AF224075 AF244740 Solanum palitans C. V. Morton 1 Bohs 2449 AF224064 AF244739 Solanum physalifolium Rusby var. 1 Bohs 2467 U47421 AF244737 nitidibaccatum 2Bitter) Edmonds Solanum pseudocapsicum L. 2 BIRM S.0870 U47422 AF244720 Solanum ptychanthum Dunal 2 RGO S-94 U47423 AF244735 Sw. 2 BIRM S.0389 L76286 AF244729 Solanum tripartitum Dunal 1 Bohs 2465 U72750 AF244738 Solanum trizygum Bitter 1 Bohs 2511 U72754 AF244745 Solanum vespertilio Aiton 2 RGO S-103 AF224070 AF244727 Solanum villosum Mill. 1 Bohs 2553 AF224066 AF244736 Solanum wallacei 2A.Gray) Parish 1 Bohs 2438 U47426 AF244741 Solanum wendlandii Hook. f. 2 BIRM S.0488 U47427 AF244731 Triguera osbeckii 2L.) Willk. 2 Jury 13742 2RNG) AF224065 AF244734 Witheringia solanacea L'Her. 1 Bohs 2416 U72755 AF244712 a DNA extracts provided by: 1 ± L. Bohs, University of Utah, Salt Lake City, UT. 2 ± R. G. Olmstead, University of Washington, Seattle, WA. 3 ± T. Mione, Central Connecticut State University, New Britain, CT b Collector and number of herbarium vouchers. Bohs vouchers are at UT, RGO vouchers at WTU. BIRM samples bear the seed accession number of the University of Birmingham Solanaceae collection 38 L. Bohs and R. G. Olmstead: A reassessment of Normania and Triguera 2Solanaceae)

Doyle and Doyle 21987). Where sampling coincided heuristicsearches using the same parameters, but with the previous studies cited above, the same with Solanum 2excluding Normania and Triguera) DNA extracts were used. PCR ampli®cation of the constrained to monophyly were conducted to ndhF region was accomplished using the methods examine how much less parsimonious it would be described in Bohs and Olmstead 21997). Ampli®ca- to exclude those taxa. Bootstrap analyses were tion of the ITS region used primers ITS 4 and ITS performed with 500 replicates using the heuristic leu1 25¢-GTCCACTGAACCTTATCATTTAG-3¢) search option, TBR, and MulTrees, with Maxtrees in 25 ll reactions containing the following: 12.25 ll set to 1000. Initial runs of the maximum likelihood water, 1.25 ll each 10 lM primer, 4.15 ll Perkin analyses used one of the most parsimonious trees

Elmer 10X bu€er containing 15 mM MgCl2, 2.5 ll from the parsimony analyses and varied the 2.5 mM dNTPs, 1.25 ll glycerol, 1.25 ll DMSO, substitution model, base frequencies, and among- 0.1 ll AmpliTaq. The PCR program used for ITS site rate variation using the options supplied in ampli®cation was 97° C for 2 min followed by 30 PAUP*. For each combination of parameters, a cycles of 97° C for 1 min, 50° C for 1 min, 72° C likelihood score was computed and scores were for 45 sec, with a 3 sec extension per cycle, and a compared using likelihood ratio tests. For all data single cycle of 72° C for 7 min. Ampli®ed products sets, the best likelihood score was obtained using a were cleaned using QiaQuick spin columns 2Qia- general-time-reversible model with rate heteroge- gen, Inc., Valencia, CA) and were sequenced on an neity and with the base frequencies, rate matrix ABI automated sequencer. Sequencing of ndhF parameters, proportion of invariable sites, and used the eight to ten primers described in Bohs and shape of the gamma distribution estimated from Olmstead 21997). Sequencing of ITS used primers the data using maximum likelihood. These esti- ITS 4 and ITS 5 of White et al. 21990); ITS 2 and mated values were then used to compute a like- ITS 3 were also used in some taxa. lihood tree and score for each data set using 1000 Sequence data were edited and contigs con- replicates of quartet puzzling. structed using the computer program Sequencher In order to determine whether ITS sequence 2Gene Codes Corp.). After a consensus sequence divergence could be used to calculate the approx- was obtained from all primer data, it was aligned imate age of Normania in Macaronesia, likelihood by eye to a template sequence [Nicotiana tabacum scores were computed on one of the ITS trees from L. for ndhF, Solanum diploconos 2Mart.) Bohs for the parsimony analysis using the Enforce Molec- ITS]. Base changes relative to the template ular Clock option. The likelihood scores were sequence were double-checked against the chroma- compared by likelihood ratio tests to those tograms, and alignments were adjusted by eye using obtained without invoking the clock option. the program Se-Al 2Rambaut 1996). Due to ambiguities in alignment of the ITS data, 78 Results characters were excluded from subsequent analyses of the individual ITS and combined ITS and ndhF The ndhF sequences obtained for all taxa data sets. All new sequences obtained in this study except S. wendlandii and L. heteroclita were were submitted to GenBank 2Table 1), and the 2086 base pairs long, corresponding to posi- complete data set and trees depicted in Figs. 3±5 tions 24 through 2109 in the tobacco ndhF have been submitted to TreeBASE. sequence. Solanum wendlandii had a 33 bp A partition homogeneity test with 1000 repli- insertion and L. heteroclita a 15 bp insertion cates was performed on the combined data set between positions 1476 and 1477 in the ndhF using PAUP* 4.0b2a 2Swo€ord 1999) to determine sequence. All sequences were easily alignable if the data sets exhibited signi®cant heterogeneity. by eye. Parsimony and maximum likelihood analyses were conducted on the individual and combined data The ndhF data set contained 341 variable sets using PAUP* 4.0b2a. The parsimony analyses characters, of which 175 were parsimony used the heuristicsearchalgorithm with the TBR informative. Pairwise sequence divergence cal- and MulTrees options, equal weights for all culated using the Kimura 2-parameter model nucleotide positions, gaps treated as missing data, ranged from 0.24% for the closely related and 100 random-order entry replicates. Additional taxon pairs S. villosum vs. S. ptychanthum and L. Bohs and R. G. Olmstead: A reassessment of Normania and Triguera 2Solanaceae) 39

Fig. 3. One of 240 most parsimonious trees of 499 steps from the parsimony analysis of the ndhFdata. Numbers above branches are branch lengths; numbers below branches are bootstrap values. Dashed lines indicate branches that collapse in the strict consensus tree. Taxa assigned to Solanum subgenera Leptostemonum, Minon, Potatoe,andSolanum as de®ned by D'Arcy 21972, 1991) are indicated. Arrows mark species endemic to Macaronesia 40 L. Bohs and R. G. Olmstead: A reassessment of Normania and Triguera 2Solanaceae)

Fig. 4. One of 22 most parsimonious trees of 927 steps from the parsimony analysis of the ITS data matrix. Numbers above branches are branch lengths; numbers below branches are bootstrap values. Dashed lines indicate branches that collapse in the strict consensus tree. Taxa assigned to Solanum subgenera Leptostemonum, Minon, Potatoe,andSolanum as de®ned by D'Arcy 21972, 1991) are indicated. Arrows mark species endemic to Macaronesia L. Bohs and R. G. Olmstead: A reassessment of Normania and Triguera 2Solanaceae) 41

Fig. 5. One of 16 most parsimonious trees of 1444 steps from the parsimony analysis of the combined ndhFand ITS data matrix. Numbers above branches are branch lengths; numbers below branches are bootstrap values. Dashed lines indicate branches that collapse in the strict consensus tree. Taxa assigned to Solanum subgenera Leptostemonum, Minon, Potatoe,andSolanum as de®ned b D'Arcy 21972, 1991) are indicated. Arrows mark species endemic to Macaronesia 42 L. Bohs and R. G. Olmstead: A reassessment of Normania and Triguera 2Solanaceae)

S. tripartitum vs. S. palitans to 3.4% between 16.3% in S. trizygum vs. Physalis alkekengi. S. candidum and Lycianthes heteroclita. Pair- Pairwise sequence divergence was 7.62% wise sequence divergence between Normania between Normania and Triguera. and Triguera in ndhF was 0.72% using this A parsimony analysis of the ITS data model. resulted in 22 most parsimonious trees of 927 Parsimony analysis of the ndhF data steps, with a CI 2excluding uninformative resulted in 240 most parsimonious trees of characters) of 0.368 and RI of 0.534. Normania 499 steps, with a CI 2excluding uninformative and Triguera form a clade nested within characters) of 0.656 and RI of 0.822. Normania Solanum in all most parsimonious trees triphylla and Triguera osbeckii form a clade 2Fig. 4). The Normania/Triguera clade emerges and in all most parsimonious trees. This as a basal lineage within Solanum, but, as in clade is nested within a large, well-supported the ndhF trees, its relationships to other clade that includes the rest of Solanum 2Fig. 3). Solanum subgroups are unresolved. The con- The relationship of Normania and Triguera strained search required three additional steps with respect to the sampled Solanum species is to ®nd a monophyletic Solanum, excluding unresolved, however, because this clade forms Normania and Triguera. Analysis of the data part of a polytomy at the base of Solanum. The using maximum likelihood resulted in a tree best trees obtained by the constrained search in topology 2not shown) much di€erent from the which Normania and Triguera are excluded parsimony trees. Normania and Triguera from Solanum are one step longer and place belong to the same clade, but they form a Normania and Triguera as the sister group to polytomy with S. luteoalbum. Support for this Solanum. A maximum likelihood analysis of grouping is low 217% of the quartet puzzling the ndhF data resulted in a topology 2not replicates). Furthermore, there is no support shown) congruent with that of the strict for the relationship of this clade to other consensus tree from the parsimony analysis, subgroups of Solanum. The incongruence in in which Normania and Triguera form a well- topologies resulting from the parsimony and supported clade nested within Solanum, but maximum likelihood analyses indicates sub- relationships of this clade to other Solanum stantial di€erences in branch lengths through- subgroups are unresolved. out the tree, and this is con®rmed by the The length of the aligned ITS sequences is likelihood ratio tests comparing scores with 710 nucleotides, including 9 bp of the 18S and without a molecular clock. The hypothesis rDNA gene at the 5¢ end of the sequence, of rate homogeneity was rejected in all 14 bp of the 26S gene at the 3¢ end of the comparisons using the HKY and GTR models sequence, and 164 bp of the 5.8S rDNA gene with a variety of parameters. These results intercalated between ITS-1 and ITS-2. The indicate that ITS sequence divergence values length of ITS-1 ranged from 210 bp in cannot be used to estimate the age of S. palitans to 257 bp in S. mammosum. The divergence of Normania and Triguera. length of ITS-2 ranged from 206 bp in Resampling the data partitions with 1000 S. candidum, S. physalifolium, S. ptychanthum, replicates of the partition homogeneity test S. villosum, and L. heteroclita to 226 bp in gave a value of p ˆ 0.10, indicating that the Capsicum baccatum and Jaltomata procum- data sets were not signi®cantly di€erent from bens. Of the 710 nucleotides, 78 were excluded random partitions of the combined data set. from analyses due to alignment ambiguities. The ndhF and ITS data were then analyzed Of the remaining 632 characters, 255 were together, giving a matrix of 2751 characters. variable and 170 were parsimony informative. Of these, 596 were variable and 345 were Pairwise sequence divergence calculated using parsimony informative. A parsimony analysis the Kimura 2-parameter model ranged from found 16 shortest trees of 1444 steps, with a CI 0.65% in S. aviculare vs. S. laciniatum to 2excluding uninformative characters) of 0.441 L. Bohs and R. G. Olmstead: A reassessment of Normania and Triguera 2Solanaceae) 43 and an RI of 0.625. Normania and Triguera is estimated at 12±14 myr and that of Madeira form a well-supported clade, as in the separate at 5 myr 2Francisco-Ortega et al. 1996). Thus, analyses 2Fig. 5). The Normania/Triguera Normania could have been present in Madeira clade is sister to a clade consisting mainly of and the Canary Islands since the Pliocene. The members of the Solanum subgenera Solanum sister relationship of Normania and Triguera is and Potatoe. Although this relationship ap- an example of a connection between endemic pears in all most parsimonious trees, it has Macaronesian taxa and a mainland group little support 217% of the bootstrap replicates). from Iberia and northwestern Africa. Other Likewise, this relationship was not supported examples of this distribution pattern that have in the maximum likelihood analysis 2results been investigated using molecular data include not shown). Though the Normania/Triguera Argyranthemum and the Asteriscus alliance clade appeared in 94% of the quartet puzzling 2Asteraceae; Francisco-Ortega et al. 1995, replicates, it emerges within Solanum near the 1997, 1999), Echium 2Boraginaceae; BoÈ hle base of the clade with little support for a close et al. 1996), and Ixanthus 2Gentianaceae; Thiv relationship with any other subgroup in the et al. 1999). The endemicstatus of the genus. The constrained parsimony search Normania species, their morphological dis- required three additional steps to ®nd a tinctness, and their occurrence in moist laurel monophyletic Solanum s.s. forest may indicate that they are Tertiary relicts, rather than relatively recent arrivals to the archipelago via long-distance dispersal. Discussion Unfortunately, the data at hand for Normania Three main conclusions can be drawn from and Triguera are not sucient to discriminate this study: 1) Normania and Triguera are between these two hypotheses. Substantial closely related, 2) Normania and Triguera molecular divergence has occurred between clearly belong to a well-supported clade along Normania and Triguera in both ndhF and ITS. with Solanum and are best considered as However, no good calibrations exist for species of Solanum, and 3) heterandry 2unequal determining the rate of ndhF sequence evolu- anthers within a single ¯ower) has evolved at tion in Solanaceae, and rates of ndhF diver- least twice in the endemic Macaronesian gence for other taxa are unknown. Solanaceae. Each of these points is discussed Approximate rates of ITS sequence divergence in more detail below. have been estimated for other taxonomic Normania and Triguera are closely rela- groups 2e.g. Suh et al. 1993; Sang et al. 1994, ted. Sequence data from both genes support 1995; BoÈ hle et al. 1996), but the rates vary the close relationship between these two taxa, considerably among taxa, and rejection of a as suggested by previous authors 2Lowe 1872, molecular clock assumption for the Solanaceae Francisco-Ortega et al. 1993). Macro- and ITS data set means that these values cannot be micromorphological characters shared by both used reliably in formulating a hypothesis of the taxa include somewhat zygomorphic corollas, age of the Normania/Triguera split. Further- large leafy calyces, horned anthers, pollen more, Normania and Triguera form an isolated colpi joined at the poles, and cells of the seed clade without obvious relationships to other coat with radially extended walls. Solanum subgroups, so it cannot be ascer- The ¯ora of Macaronesia has strong tained whether this clade has its closest Mediterranean anities and in part may relatives among other Old World taxa and represent a relict from a formerly widely when it may have diverged from its closest distributed moist forest ¯ora of Tertiary age living relatives. 2e.g. Bramwell 1976, Sunding 1979, Francisco- Normania and Triguera are included within Ortega et al. 1997, Helfgott et al. 2000). The Solanum. All analyses of the nuclear and maximum age of Tenerife and Gran Canaria chloroplast data sets, singly and in combina- 44 L. Bohs and R. G. Olmstead: A reassessment of Normania and Triguera 2Solanaceae) tion, result in the Normania/Triguera clade ºVerbascum osbeckii L., Sp. Pl. 1: 179. 1753. being nested within Solanum. Currently Sola- Type: Ex Hispania, Osbeck 2lectotype, num is broadly de®ned, and several segregate S-LINN G-6305; Micro®che IDC genera which have recently been found to be 87.17). Designated by Hansen & Hansen included within the Solanum clade have been 21973). subsumed within it [e.g. Lycopersicon 2Spooner ºTriguera osbeckii 2L.) Willk. in Willk. & et al. 1993), Cyphomandra 2Bohs 1995)]. In Lange, Prodr. ¯. hispan. 2: 524. 1870. accordance with this taxonomic concept, we ºFontqueriella osbeckii 2L.) Rothm. in Font- recommend that the use of the generic names Quer & Rothm., Brote ria 36: 151. 1940. Normania and Triguera be abandoned and ?=Triguera ambrosiaca Cav., Diss. 2, p. 2 and their species transferred to Solanum. Names tab. A. 1786. Type: Southern Spain. In are available in Solanum for both Normania argillaceis Carmona, Hispalis, Co rdoba, species, but a new name is needed for the et per totam fere inferiorem Baeticam, transfer of Triguera osbeckii to Solanum. D. de Trigueros s.n. 2holotype, MA; Synonymy of the three species is given below: isotypes, P). ?=Triguera baccata J. F. Gmel., Syst. nat. Solanum nava Webb & Berthel., Hist. nat. Iles 221): 338. 1791. Type: unknown. Canaries 2. 323): 123. 1845; tab. 174. 1849. Type: Canary Islands. Caidero de CorunÄ a, The new name, Solanum herculeum,is montium Saucillo, supra vicum Tentenigua- taken from Herculeum Fretum, a classical dam, Herbarium Webbianum s.n., name for the Strait of Gibraltar. Solanum Solanaceae no. 9 2lectotype, FI). Designated herculeum occurs both north and south of this by Leo n et al., Vieraea 13: 21. 1984. strait. Triguera ambrosiaca and T. baccata are =Solanum nava var. undulatidentatum Bitter, regarded as synonyms in accordance with Repert. Spec. Nov. Regni Veg. 11: 253. Hansen and Hansen 21973). The speci®c status 1912. Type: In silva Teneri€e, Agua of Triguera inodora is in doubt 2Hansen and Garcia, Webb 44 2lectotype, W, Hansen 1973), and is not considered here. #288851). Designated by Francisco-Or- Likewise, Lowe 21872) and Francisco-Ortega tega et al. 21993). et al. 21993) have suggested that Normania ºNormania nava 2Webb & Berthel.) Franc.- nava and N. triphylla are conspeci®c. A critical Ort. & R. N. Lester, Pl. Syst. Evol. 185: examination of species boundaries among the 202. 1993. taxa of Triguera and Normania is beyond the Solanum trisectum Dunal in DC. Prodr. 1321): scope of the current paper. 36. 1852. Non S. triphyllum Vell. Fl. Flum. 2: Relationships of the Normania/Triguera 120. 1827. clade to other groups within Solanum are ºNycterium triphyllum Lowe, Trans. Cam- obscure. Though placed near the tuber-bear- bridge Philos. Soc. 6: 536. 1838. Syn- ing by previous authors [Dunal, types: Madeira, S. Vicente below the 1852 2as subsection Tuberarium); D'Arcy, Gingeiras, on the roadside to the Paul, 1972 and Child, 1990 2as subgenus Potatoe)], ca. 1000 ft, July 1837, Lemann 1030 molecular data indicate that this clade is 2BM, G-DC); Madeira, in the east near probably not closely allied with the potatoes Portella, Lippold 2W). and their relatives 2Figs. 3±5). Parsimony ºNormania triphylla 2Lowe) Lowe, Man. ¯. analysis of the combined molecular data set Madeira 221): 87. 1872. resulted in an association of the Normania/ Triguera clade with a clade consisting mainly Solanum herculeum Bohs, nom. nov. Non of members of Solanum subgenera Potatoe S. osbeckii Dunal in DC. Prodr. 1321): 179. and Solanum 2Fig. 5), but this relationship is 1852. not well supported. No obvious macromor- L. Bohs and R. G. Olmstead: A reassessment of Normania and Triguera 2Solanaceae) 45 phological synapomorphies de®ne this larger facilitates outcrossing in other Solanum species clade, and Francisco-Ortega et al. 21993) with unequal anthers. concluded that micromorphological features Solanum vespertilio was included in the of the pollen grains and seed coat cell wall molecular analyses, and is indeed nested within thickenings were quite di€erent between the spiny Solanum clade 2Figs. 3±5). The Normania/Triguera and members of Solanum results do not support a close evolutionary subgenus Potatoe. To date, the molecular and relationship between Normania/Triguera and morphological data suggest that the Norma- S. vespertilio, indicating that ¯oral zygomor- nia/Triguera clade may be a relatively basal phy and heterandry have evolved more than lineage of Solanum without extant close once within the Macaronesian Solanum species. relatives. Among the sampled taxa, S. vespertilio appears Solanum section Normania 2Lowe) Bitter to be most closely related to S. melongena and was set up by Bitter 21912) to accommodate S. macrocarpon of the S. incanum group sensu the two Macaronesian taxa S. nava and Whalen [1984; sections Melongena 2Mill.) S. trisectum. It is useful to retain this section, Dunal or Andromonoecum Bitter], an Old but amend its de®nition to include the former World group with actinomorphic ¯owers and genus Triguera. As such, it would be com- equal anthers. Other molecular results prised of three species, Solanum nava, 2Olmstead and Palmer 1997; Bohs, unpub- S. trisectum, and S. herculeum. lished) further indicate that S. vespertilio is not Heterandry has evolved at least twice in the closely related to the New World members of endemic Macaronesian species of Solanum. section Nycterium, despite their similar zygo- Aside from S. nava and S. trisectum, two other morphic¯owers with unequal anthers. Solanum species, S. liddii and S. vespertilio, are The inclusion of the genera Normania and endemicto Macaronesia. Both belong to Triguera within Solanum in no way diminishes section Nycterium in the spiny subgenus the evolutionary and biogeographicimportance Leptostemonum and are obviously closely of these taxa and their priority for conservation. related. Although S. liddii and S. vespertilio Regardless of their taxonomicdesignation, they possess spines and stellate hairs typical of the should be the focus of intensive e€orts to locate other members of the subgenus, they are new populations and conserve existing ones. We unusual in having highly zygomorphic¯owers are in complete agreement with Francisco- with unequal anthers. Heterandry 2the pre- Ortega et al. 21993), who advocate preservation sence of unequal anthers in a single ¯ower) is of the Macaronesian laurel forest habitat as well found in several sections of subgenus Lepto- as ex situ conservation measures for S. trisectum stemonum [e.g. sect. Nycterium 2Vent.) Dunal, and S. nava. Although not endangered at sect. Androceras 2Nutt.) Marzell, sect. Ani- present, e€orts should be made to monitor and santherum Bitter, sect. Mondolichopus Bitter, preserve populations of S. herculeum and to sect. Aculeigerum Seithe] as well as in non- introduce this species into seed collections and/ spiny groups [e.g. sect. Jasminosolanum 2Bitter) or botanical gardens where its biology and Seithe, sect. Geminata 2G. Don) Walp., sect. morphology can be studied in detail. Close Lycopersicum 2Mill.) Wettst.] and has appar- comparison of living material of these taxa may ently evolved multiple times in the genus reveal other morphological similarities and will Solanum. The functional signi®cance of heter- help to resolve species boundaries in the group. andry in pollination was studied in S. ros- Field studies of the pollination biology of the tratum Dun. of section Androceras by Bowers Macaronesian species of Solanum as well as 21975), who concluded that outcrossing was other taxa of Solanum with heteromorphic promoted by heteromorphicanthers in combi- anthers are needed to understand the possible nation with enantiostyly 2asymmetrical style adaptive signi®cance of this striking conver- placement). It is not known if heterandry gence in ¯oral morphology in disparate clades 46 L. Bohs and R. G. Olmstead: A reassessment of Normania and Triguera 2Solanaceae) within the genus. In addition, it would be useful woody habit in Echium 2Boraginaceae). Proc. to know what signi®cance the horned anthers of Natl. Acad. Sci. USA 93: 11740±11745. S. nava, S. herculeum, and S. trisectum might Bohs L. 21995) Transfer of Cyphomandra 2Solana- have in attracting or manipulating pollinators. ceae) and its species to Solanum. Taxon 44: 583± Investigations of ¯oral visitors to S. nava and 587. S. trisectum in the Canary Islands are even more Bohs L., Olmstead R. G. 21997) Phylogenetic relationships in Solanum 2Solanaceae) based on urgent given the rare status of these plants in ndhF sequences. Syst. Bot. 22: 5±17. their native habitats. Certainly an e€ective Bohs L., Olmstead R. G. 21999) Solanum phylogeny conservation plan should consider aspects of inferred from chloroplast DNA sequence data. the reproductive biology of these plants such as In: Nee M., Symon D. E., Lester R. N., Jessop J. their ¯oral visitors or pollinators. P. 2eds.) Solanaceae IV: advances in biology and utilization. Royal BotanicGardens, Richmond, We thank the many researchers and botanical Kew, UK, pp. 97±110. gardens that provided plant material for this study Bowers K. A. W. 21975) The pollination ecology of and the greenhouse sta€ of Duke University and Solanum rostratum 2Solanaceae). Amer. J. Bot. the University of Utah for maintaining living 62: 633±638. collections. The Instituto Nacional de Biodiversi- Bramwell D. 21976) The endemic¯ora of the Canary dad and the Organization for Tropical Studies Islands; distribution, relationships, and phyto- facilitated processing of Costa Rican collecting geography. In: Kunkel G. 2ed.) Biogeography and permits, and the Tropical Science Center, San Jose , ecology in the Canary Islands. Dr. W. Junk, The Costa Rica, granted permission to collect Solanum Hague, The Netherlands, pp. 207±240. trizygum in the Monteverde Cloud Forest Reserve. Cavanilles A. J. 21786) Monadelphiae classis Special thanks are due to A. Child, R. Lester, J. dissertationes decem 2: i-iii, pl. A. Typographia Francisco-Ortega, and J. R. Acebes Ginove s for regia, Madrid. their help in obtaining material of Normania, to Dr. Child A. 21990) A synopsis of Solanum subgenus S. Knapp for sending relevant literature, to Dr. J. Potatoe 2G. Don) D'Arcy 2Tuberarium 2Dun.) Francisco-Ortega and an anonymous reviewer for Bitter 2s.l.)). Feddes Repert. 101: 209±235. their helpful comments on the manuscript, to Dr. D'Arcy W. G. 21972) Solanaceae studies II: M. Nee for suggesting the new Solanum epithet, to typi®cation of subdivisions of Solanum. Ann. M. Whitson for providing the Physalis alkekengi Missouri Bot. Gard. 59: 262±278. ITS sequence, and to Dr. V. Stiller for help with D'Arcy W. G. 21991) The Solanaceae since 1976, German translations and Adobe Photoshop. with a review of its biogeography. In: Hawkes J. P. Reeves, S. King-Jones, and M. Johnson G., Lester R. N., Nee M., Estrada R. N. 2eds.) provided technical assistance, and P. Corneli and Solanaceae III: , chemistry, evolution. the SPUD group at the University of Utah helped Royal BotanicGardens, Richmond,Kew, UK, with the maximum likelihood analyses. The advice pp. 75±137. and expertise of our friends and colleagues in Darwin C. 21845) The voyage of H. M. S. Beagle. Solanaceae systematics is gratefully acknowledged. 2nd ed. Reprint ed., 1957. The Heritage Press, This research was supported by NSF grants DEB- New York. 9207359 and DEB-9996199 to LB and BSR- Doyle J. J., Doyle J. L. 21987) A rapid DNA 9107827 and DEB-9509804 to RGO. isolation procedure for small quantities of fresh References leaf tissue. Phytochem. Bull. 19: 11±15. Dunal M. F. 21852) Solanaceae. In: DeCandolle A. Bentham G. 21876) Solanaceae. In: Bentham G., P. 2ed.) Prodromus systematis naturalis regni Hooker J. D. 2eds.) Genera plantarum 222). vegetabilis 1321). Victoris Masson, Paris, pp. 1± Lovell Reeve and Co., London, pp. 882±913. 690. Bitter G. 21912) Solana nova vel minus cognita. IV. Farr E. R., Leussink J. A., Sta¯eu F. A. 2eds.) Repert. Spec. Nov. Regni Veg. 11: 241±260. 21979) Index nominum genericorum 2planta- BoÈ hle U. R., Hilger H. H., Martin W. F. 21996) rum). Vol. 3. Dr. W. Junk, The Hague, The Island colonization and evolution of the insular Netherlands. L. Bohs and R. G. Olmstead: A reassessment of Normania and Triguera 2Solanaceae) 47

Francisco-Ortega J., Hawkes J. G., Lester R. N., Lowe R. T. 21838) Novitiae ¯orae maderensis, or Acebes-Ginove s J. R. 21993) Normania,an notes and gleanings of Maderan botany. Trans. endemic Macaronesian genus distinct from Cambridge Philos. Soc. 6: 523±551. Solanum 2Solanaceae). Plant Syst. Evol. 185: Lowe R. T. 21872) A manual ¯ora of Madeira and 189±205. the adjacent islands of Porto Santo and the Francisco-Ortega J., Jansen R. K., Crawford D. J., Desertas. Vol. 2, part 1. John van Voorst, Santos-Guerra A. 21995) Chloroplast DNA London. evidence for intergeneric relationships of the Miers J. 21849a) On the genus Triguera of Macaronesian endemic genus Argyranthemum Cavanilles. Hooker's J. Bot. Kew Gard. Misc. 2Asteraceae). Syst. Bot. 20: 413±422. 1: 65±67. Francisco-Ortega J., Jansen R. K., Santos-Guerra Miers J. 21849b) Observations upon several genera A. 21996) Chloroplast DNA evidence of coloni- hitherto placed in Solanaceae, and upon others zation, adaptive radiation, and hybridization in intermediate between that family and the Scro- the evolution of the Macaronesian ¯ora. Proc. phulariaceae. Ann. Mag. Nat. Hist. ser. 2, 3: Natl. Acad. Sci. USA 93: 4085±4090. 161±183. Francisco-Ortega J., Santos-Guerra A., Hines A., Olmstead R. G., Palmer J. D. 21992) A chloroplast Jansen R. K. 21997) Molecular evidence for a DNA phylogeny of the Solanaceae: subfamilial Mediterranean origin of the Macaronesian relationships and character evolution. Ann. endemicgenus Argyranthemum 2Asteraceae). Missouri Bot. Gard. 79: 346±360. Amer. J. Bot. 84: 1595±1613. Olmstead R. G., Palmer J. D. 21997) Implications Francisco-Ortega J., Goertzen L. R., Santos- for the phylogeny, classi®cation, and biogeogra- Guerra A., Benabid A., Jansen R. K. 21999) phy of Solanum from cpDNA restriction site Molecular systematics of the Asteriscus alliance variation. Syst. Bot. 22: 19±29. 2Asteraceae: Inuleae) I. Evidence from the Olmstead R. G., Sweere J. A., Spangler R. E., Bohs internal transcribed spacers of nuclear ribosomal L., Palmer J. D. 21999) Phylogeny and provi- DNA. Syst. Bot. 24: 249±266. sional classi®cation of the Solanaceae based on Gmelin J. F. 21791) Systema naturae. Ed. 13, 221): chloroplast DNA. In: Nee M., Symon D. E., 338. Lester R. N., Jessop J. P. 2eds.) Solanaceae IV: Hansen A., Hansen C. 21973) Verbascum osbeckii advances in biology and utilization. Royal L. and the Triguera-species 2Solanaceae). A BotanicGardens, Richmond, Kew, UK, pp. revision. Lagascalia 3: 183±193. 111±137. Hawkes J. G. 21972) Triguera. Solanaceae. In: Poiret J. L. M. 21808) In: Lamarck J. B., Tutin T. G., Heywood V. H., Burges N. A., Encyclope die me thodique. Botanique 8: 99. Moore D. M., Valentine D. H., Walters S. M., Rambaut A. 21996) Se-Al. Sequence alignment Webb D. A. 2eds.) Flora Europaea. Cambridge editor. Version 1.0 alpha 1. Distributed on the University Press, London, p. 200. web at http: //evolve.zoo.ox.ac.uk/Se-Al/ Helfgott D. M., Francisco-Ortega J., Santos- Se-Al.html. Guerra A., Jansen R. K., Simpson B. B. 22000) Roemer J. J., Schultes J. A. 21819). Systema Biogeography and breeding system evolution of vegetabilium 4: 698. the woody Bencomia alliance 2Rosaceae) in Sang T., Crawford D. J., Kim S.-C., Stuessy T. F. Macaronesia based on ITS sequence data. Syst. 21994) Radiation of the endemicgenus Dendro- Bot. 25: 82±97. seris 2Asteraceae) on the Juan Fernandez Lester R. N., Francisco-Ortega J., Al-Ani M. Islands: evidence from sequences of the ITS 21999) Convergent evolution of heterandry 2un- regions of nuclear ribosomal DNA. Amer. equal stamens) in Solanum, proved by spermo- J. Bot. 81: 1494±1501. derm SEM. In: Nee M., Symon D. E., Lester R. Sang T., Crawford D. J., Stuessy T. F., Silva O. M. N., Jessop J. P. 2eds.) Solanaceae IV: advances in 21995) ITS sequences and the phylogeny of the biology and utilization. Royal BotanicGardens, genus Robinsonia 2Asteraceae). Syst. Bot. 20: Kew, UK, pp. 51±69. 55±64. Linnaeus C. 21753) Species plantarum. Facsimile of Spooner D. M., Anderson G. J., Jansen R. K. the ®rst edition, 1957. The Ray Society, London. 21993) Chloroplast DNA evidence for the inter- 48 L. Bohs and R. G. Olmstead: A reassessment of Normania and Triguera 2Solanaceae)

relationships of tomatoes, potatoes, and pepinos Phytographia Canariensis. Vol. 3, section 3. 2Solanaceae). Amer. J. Bot. 80: 676±688. Paris, p. 123 and tab. 174. Sprengel K. 21825) Systema vegetabilium 1: 661. Whalen M. D. 21984) Conspectus of species groups Suh Y., Thien L. B., Reeve H. E., Zimmer E. A. in Solanum subgenus Leptostemonum. Gentes 21993) Molecular evolution and phylogenetic Herb. 12: 179±282. implications of internal transcribed spacer White T. J., Bruns T., Lee S., Taylor J. 21990) sequences of ribosomal DNA in Winteraceae. Ampli®cation and direct sequencing of fungal Amer. J. Bot. 80: 1042±1055. ribosomal RNA genes for phylogenetics. In: Sunding P. 21979) Origins of the Macaronesian Innis M., Gelfand D., Sninksy J., White T. 2eds.) ¯ora. In: Bramwell D. 2ed.) Plants and islands. PCR protocols: a guide to methods and applica- Academic Press, London, pp. 13±40. tions. Academic Press, San Diego, California, Swo€ord D. 21999) PAUP*. Version 4.0b2. Sinauer pp. 315±322. Associates, Sunderland, Massachusetts. Willkomm M. 21870) Solanaceae. In: Willkomm Thiv M., Struwe L., Kadereit J. W. 21999) The M., Lange J. M. C. L. 2eds.) Prodromus ¯orae phylogeneticrelationships and evolution of the hispanicae 2: 524. Canarian laurel forest endemic Ixanthus visco- sus 2Aiton) Griseb. 2Gentianaceae): evidence Addresses of the authors: Lynn Bohs 2bohs@ from matK and ITS sequences, and ¯oral biology.utah.edu), Department of Biology, Uni- morphology and anatomy. Plant Syst. Evol. versity of Utah, Salt Lake City, UT 84112, USA. 218: 299±317. Richard G. Olmstead, Department of Botany, Webb P. B., Berthelot S. 21845) Histoire University of Washington, Seattle, WA 98195, naturelle des Iles Canaries. DeuxieÁ me partie. USA.