Phylogenetics of Tribe Anthocercideae (Solanaceae) Based on ndhF and trnL/F Sequence Data

Tribe Anthocercideae (Solanaceae) is an Australian endemic group comprising 31 species in seven genera. Recent phylogenetic work has placed the Anthocercideae sister to Nicotiana. Two chloroplast DNA regions, ndhF and trnL/ F, were analyzed and the phylogeny was used to test the tribe’s monophyly, discover relationships within the tribe, and make inferences on character evolution and biogeography. The relationship between Nicotiana, Symonanthus, and the rest of tribe Anthocercideae is unresolved. Anthocercis, Anthotroche, Grammosolen, and Symonanthus are found to be monophyletic, while Cyphanthera and Duboisia are not (Crenidium is monotypic). Several characters were inferred to be derived within the Anthocercideae, including unilocular stamens with semicircular slits, ebracteolate flowers, and baccate fruits. Ancient colonization occurred in southwestern Australia followed by several radiation events eastward. Tribe Anthocercideae as currently circumscribed comprises 31 species in seven genera: Anthocercis (10 spp.), Anthotroche (4 spp.), Crenidium (1 sp.), Cyphanthera (9 spp.), Duboisia (3 spp.), Grammosolen (2 spp.), and Symonanthus (2 spp.). All members are endemic to Australia, with the exception of Duboisia myoporoides, which is also found in New Caledonia. The region of greatest diversity for the group occurs in southwestern Australia, and most taxa are confined to the southern half of the continent. The tribe is characterized by a suite of morphological characters: woody habit, flowers with non-accrescent calyx, inflexo-valvate aestivation of the corolla lobes, a short, relatively broad actinomorphic corolla tube, extrorsely dehiscing stamens inserted low in the corolla tube, and an oblong to ellipsoid, slightly curved seed with reticulate testa in which is held a terete, slightly curved embryo (Haegi 1986). Solanaceae tribe Anthocercideae was established by Don (1838) to contain Anthocercis and Duboisia, the only two of the present seven genera described at that time. Miers (1849) established tribe Duboiseae to include Anthotroche and Cyphanthera, along with Anthocercis and Duboisia. Bentham (1846, 1868) transferred Anthocercis and Duboisia to the Scrophulariaceae based on the presence of four didynamous stamens. Baehni (1946) resurrected the Anthocercideae, but included only Anthocercis (along with four genera now known to be unrelated), placing Duboisia in the Salpiglossideae and leaving Anthotroche and Isandra (5 Symonanthus) incertae sedis. He did not recognize Cyphanthera. Haegi (1979, 1981, 1986, 1991) provides the most complete treatment of the Anthocercideae and established its present taxonomy, describing two new genera, Crenidium and Grammosolen (Haegi 1981) and providing a correct name for Symonanthus (Isandra was a later homonym). Recent morphological and phytochemical studies have suggested the monophyly of the Anthocercideae (Haegi 1979, 1981, 1986; Purdie et al. 1982; Tétényi 1987; Knapp et al. 2000). A molecular phylogenetic study of the Solanaceae based on chloroplast DNA (cpDNA) restriction site variation placed the Anthocercideae sister to Nicotiana (Olmstead and Palmer 1992). This result was confirmed by additional studies of cpDNA restriction site, ndhF and rbcL variation (Olmstead and Sweere 1994; Olmstead et al. 1999). These studies, however, only sampled four species among the Anthocercideae: Anthocercis viscosa, Cyphanthera anthocercidea, Duboisia myoporoides, and Grammosolen dixonii. A more rigorous molecular study sampling across all genera is needed to produce a phylogenetic framework from which we can draw inferences about the group’s evolutionary history. The current study includes extensive sampling within the tribe, and the phylogenies generated from molecular data are used to: (1) test the monophyly of tribe Anthocercideae, (2) elucidate relationships within the tribe, (3) infer morphological character evolution, and (4) comment on available biogeographic data. The reconstructed phylogenies presented here are based on sequence variation from two cpDNA markers, ndhF and trnL/F. ndhF is a gene that encodes a subunit of the NADH dehydrogenase complex. It exhibits almost twice the average substitution rate of rbcL (Sugiura 1989; Olmstead and Sweere 1994; Soltis and Soltis 1998), and has been used previously in determining phylogenetic relationships in the Solanaceae (Olmstead and Sweere 1994; Bohs and Olmstead 1997, 2001). The trnL/F region, also used previously in phylogenetic analyses of other genera in Solanaceae (e.g., Fukuda et al. 2001), includes an intron and spacer flanking the 39 exon of the trnL gene. Because it is mostly non-coding, 610 [Volume 28 SYSTEMATIC BOTANY TABLE 1. Voucher and GenBank accession data for taxa examined in this study. For each voucher GenBank accession numbers are given in the following sequence: ndhF, trnL/F. Anthocercis angustifolia F. Muell., R. G. Olmstead 94-05 (WTU) AY098704, AY098671. A. gracilis Benth., H. Stace s. n.; herbarium unknown, AY098705, AY098672. A. ilicifolia Hook., H. Stace s. n. (UWA) AY098706, AY098673. A. intricata F. Muell., H. Stace s. n. (KPBG) AY098707, AY098674. A. littorea Labill., H. Stace s. n.; herbarium unknown, AY098708, AY098675. A. sylvicola Macfar. & Ward., T. Middleton s. n. (PERTH) AY098709, AY098676. A. viscosa R. Br., D. Symon 14835 (AD) U08914, AY098677. Anthotroche blackii F. Muell., H. Stace s. n. (KPBG) AY098711, AY098678. A. myoporoides C.A. Gardner, H. Stace s. n. (KPBG) AY098810, AY098679. A. pannosa Endl., H. Stace s. n. (KPBG) AY098712, AY098680. A. walcottii F. Muell., D. R. & B. Bellairs 2035 (PERTH) AY098713, AY098681. Crenidium spinescens Haegi, B. J. Lepschi & T. R. Lally 1672 (CANB) AY098714, AY098682. Cyphanthera albicans (A. Cunn.) Miers, B. J. Lepschi & T. R. Lally 1722 (CANB) AY098715, AY098683. C. anthocercidea (F. Muell.) Haegi, L. Haegi 1456 (AD) AY098716, AY098684. C. microphylla Miers, B.J. Lepschi 2170 (PERTH) AY098717, AY098685. C. myosotidea (F. Muell.) Haegi, Alcock 9117 (AD) AY098686. C. odgersii (F. Muell.) Haegi, Chinnock 3100 (AD) AY098718, AY098687. C. racemosa (F. Muell.) Haegi, L. Haegi 1959 (AD) AY098688. Duboisia leichhardtii (F. Muell.) F. Muell., L. Haegi 2056 (AD) AY098719, AY098689. D. myoporoides R. Br., D. Symon 14832 (AD) AY098720, AY098690. Grammosolen dixonii (Muell. & Tate) Haegi, D. Symon 14833 (AD) AY098721, AY098691. G. truncatus (Ising) Haegi, Canty 2429 (AD) AY098722, AY098692. Symonanthus aromaticus (C.A. Gardner) Haegi, J. McKinney s. n.; herbarium unknown, AY098723, AY098693. S. bancroftii (F. Muell.) Haegi, H. Stace s. n. (KPBG) AY098724, AY098694. Jaltomata procumbens (Cav.) J.L. Gentry, R. G. Olmstead S-24 (WTU) U47429, AY098695. Lycium cestroides Schltdl., R. G. Olmstead S-34 (WTU) U08920, AB036578 & AB036607. Nicotiana acuminata (Graham) Hook., R. G. Olmstead S-39 (WTU) U08923, AY098696. N. attenuata Torr., R. G. Olmstead S-41 (WTU) AY098697. N. excelsior (Black) Black, R. G. Olmstead S-44 (WTU) AY098725, AY098698. N. glutinosa L., no voucher, AY098726, AY098699. N. gossei Damin., R. G. Olmstead S-48 (WTU) AY098727, AY098700. N. paniculata L., R. G. Olmstead S53 (WTU) AY098728, AY098701. N. tabacum L., no voucher, L14953, Z00044. Petunia axillaris (Lam.) Britton, Stern & Poggenb., R. G. Olmstead S-60 (WTU) U08926, AY098702. Solanum lycopersicum L., no voucher, U08921, AY0987. it exhibits a higher substitution rate than ndhF, thus it was selected with hopes that it might help to resolve relationships among closely related members of the tribe. MATERIALS AND METHODS We sampled 24 species across all seven genera in the Anthocercideae (Table 1). trnL/F was sequenced for all 24 species, whereas ndhF was only sequenced for 22 species (two species of Cyphanthera would not amplify for ndhF). Due to the hypothesized sister relationship of Nicotiana to the group (Olmstead and Palmer 1992; Olmstead et al. 1999), seven species of Nicotiana were included (N. attenuata would not amplify for ndhF). Four other species from the Solanaceae were added to broaden outgroup sampling: Jaltomata procumbens, Lycium cestroides, Petunia axillaris, and Solanum lycopersicum. Total genomic DNA was extracted from herbarium samples, provided by David Symon (Adelaide Botanic Garden), Helen Stace (U. Western Australia), and Brendan Lepschi (Australian National Herbarium, Canberra) using the CTAB micro-extraction procedure (Doyle and Doyle 1987). All extractions were purified using the QIAquick Purification Kit (QIAGEN). Primers used for ndhF and trnL/F amplification and sequencing are described in Olmstead and Sweere (1994) and Taberlet et al. (1991), respectively. Double stranded PCR products were cleaned using polyethylene glycol (PEG) DNA precipitation, and quantified by spectrophotometry. Direct DNA sequencing was accomplished in both directions using the ABI Dye Terminator Cycle Sequencing Reaction Kit (Perkin Elmer, Foster City, CA), and analyzed on an ABI 377 automated sequencer. Sequences were manually edited and assembled using Sequencher 3.0 (Gene Codes, Ann Arbor, Michigan, USA). All sequences were easily aligned by eye. Parsimony informative alignment gaps were coded as binary characters. Phylogenetic analyses were performed using PAUP* 4.0b10 (Swofford 1998) using equally-weighted parsimony. A heuristic search involving 50,000 random taxon addition replicates with tree bisection reconnection (TBR) branch swapping, Collapse, and MulTrees was conducted for each data set to search for multiple islands (Maddison 1991; Page 1993). Two heuristic searches were performed on the combined dataset: the first included only taxa that were sequenced for both ndhF and trnL/F, and the second included all taxa, with missing ndhF sequences coded as missing characters. Support for individual branches was determined using bootstrap analysis (Felsenstein 1985) involving 100 replicates with 20 random taxon additions, TBR, Collapse and MulTrees. A decay analysis (Bremer 1988; Donoghue et al. 1992) was also conducted on the combined dataset. The incongruence length difference (ILD) test was conducted to determine whether the two cpDNA regions were significantly different from random partitions of the combined data (Farris et al. 1994). This was implemented as the partition homogeneity test in PAUP* using 1,000 replicates and 1,000 random taxon additions. The data set is available on TreeBASE (study accession number 5 S899; matrix accession numbers 5 M1468 and M1477).