Selbyana 25(2): 225-238. 2005.

REORGANIZAnON OF TRIBAL AND GENERIC BOUNDARIES IN THE GLOXINIEAE (GESNERIACEAE: GESNERIOIDEAE) AND THE DESCRIPTION OF A NEW TRIBE IN THE GESNERIOIDEAE, SPHAERORRHIZEAE

ERIC H. ROALSON* School of Biological Sciences and Center for Integrated Biotechnology, Washington State University, Pullman, WA 99164-4236 USA. Email: [email protected]

JOHN K. BOGGAN AND LAURENCE E. SKOG National Museum of Natural History, Department of Botany, Smithsonian Institution, Washington, DC 20013-7012 USA.

ABSTRACT. Morphological and molecular studies in tribe Gloxinieae have led to the need to describe four new genera and one new tribe, with two historically recognized genera resurrected and three currently recognized genera submerged into other generic concepts. The new genera Gloxinella, Gloxiniopsis, Nom­ opyle, and Sphaerorrhiza include species previously treated in Gloxinia. The genus Sphaerorrhiza also is treated as a new tribe because of its distant phylogenetic relationship to the Gloxinieae. Mandirola and Seemannia have been resurrected to define monophyletic groups of species previously treated in Gloxinia. The genera Anodiscus and Koellikeria have been submerged into the new circumscription of Gloxinia to reflect phylogenetic relationships and morphological similarities among the species of these genera. The circumscription of Kohleria is here broadened to include Capanea. In all, seven generic transfers of already available names are made as well as 11 new combinations: Gloxinella lindeniana, Gloxinia erinoides, G. xanthophylla, Gloxiniopsis racemosa, Kohleria affinis, K. tigridia, Mandirola rupestris, Nomopyle dodson ii, N. peruviana, Sphaerorrhiza sarmentiana, and S. burchellii.

Key words: Gesneriaceae, Gesnerioideae, Gloxinieae, Sphaerorrhizeae

INTRODUCTION current generic placement of a taxon is clearly wrong, but there is not reasonable support for its Recent studies of phylogenetic relationships placement in the generic concepts presented in Gesneriaceae subfamily Gesnerioideae (Zim­ here, we have dealt with the taxon as incertae mer et aI. 2002) and tribe Gloxinieae (Roalson sedis. Further work will be necessary to deter­ et al. 2003, Smith et al. 2004, Roalson et aI. mine the placement of these taxa. The problems 2005, E. Roalson et al. unpubl. data) have sug­ associated with delimitation and polyphyly of gested that tribal and generic boundaries in these Phinaea (Smith et al. 2004, Roalson et aI. 2005) groups require extensive reorganization. This will be addressed in a separate publication (1. paper begins the process of reorganizing generic Boggan et aI. unpubI. data). Finally, some taxa boundaries in tribe Gloxinieae, reinstating old previously placed in the Gloxinieae, but for generic concepts for some groups, and creating which there is now good evidence that they be­ new generic names where necessary. A charac­ long elsewhere, are discussed and their classifi­ terization of the phylogenetic relationships of cation position is clarified. Among the most sig­ genera (as circumscribed here) within the Glox­ nificant results of these studies are that Gloxinia inieae is presented in FIGURE 1. The currently sensu Wiehler (1976, 1983) is a polyphyletic as­ accepted species for all genera of the tribe are semblage that requires considerable reorganiza­ enumerated below; complete synonymies for the tion, and that Gloxinia sarmentiana should not species are listed by Skog and Boggan (2005). only be excluded from the genus Gloxinia but The placement of some species is tentative, as from tribe Gloxinieae. A key to genera of the they have not been sampled in previous molec­ recircumscribed Gloxinieae and a key to Ges­ ular phylogenetic studies. The species that have nerioideae tribes with inferior or half-inferior been sampled in these phylogenetic studies are ovaries are presented. denoted with an asterisk (*) in the species lists below. Where species identity is unclear or the GLOXINIEAE FRITSCH Achimenes C.H.Persoon, Syn. PI. 2: 165. 1807 * Corresponding author. [Nov 1806], nom. cons. against Achimenes 225 226 SELBYANA Volume 25(2) 2005

Achimenes data), although currently we cannot exclude the Solenophora possibility that Achimenes is monophyletic (E. Eucodonia Roalson et al. unpubl. data). Achimenes as cur­ Smithiantha rently circumscribed is morphologically hetero­ Moussonia geneous, and if eventually shown to be paraphy­ Niphaea letic with regard to Solenophora, it may be nec­ Gloxinia essary to resurrect the genera Dicyrta Regel and Seemannia Plectopoma Hanstein to include those species of Mandirola Achimenes that may be more closely related to Goyazia Solenophora. If this is the case, Dicyrta would 1------Heppiella Kohleria likely include Achimenes brevifolia, A. obscura, Pearcea and A. misera, and Plectopoma would include Phinaeap.p A. glabrata, as Plectopoma jimbriatum (W.I. '---- Diastema vexans Hooker) Hanstein. Further study of phylogenetic ...... ---- Gloxiniopsis relationships will be needed to assess whether Diastema these generic recircumscriptions are necessary. Gloxinella Diastema G.Bentham, Bot. Voy. Sulphur 132. Monopyle Phinaea s.s. 1844 [14 Apr 1845]. TYPE SPECIES: Diaste­ ma racemiferum G.Bentham. ----Nomopyle The genus includes *Diastema affine Fritsch, FIGURE 1. Hypothesis of phylogenetic relation­ ships among genera based on the studies of Roalson *D. comiferum (DC.) G.Bentham ex Walpers, D. et al. (2005) and E. Roalson et al. (unpubl. data). eggersianum Fritsch, D. gymnoleuca Gilli, D. hispidum (DC.) Fritsch, D. kalbreyeri Fritsch, D. latiflorum Rusby, D. lehmannii Regel, D. ma­ culatum (Poeppig) G.Bentham ex Walpers, D. P.Browne 1756, and Vahl 1791 (Scrophul.). micranthum I.D.Smith, D. purpurascens Rusby, TYPE SPECIES: Achimenes coccinea (Scopo­ D. quinquevulnerum I.E.Planchon & Linden, Ii) C.H.Persoon (=A. erecta (Lam.) *D. racemiferum G.Bentham, D. rupestre Bran­ H.P.Fuchs). degee, *D. scabrum (Poeppig) G.Bentham ex The genus includes *Achimenes admirabilis Walpers, D. sodiroanum Fritsch, D. tenerrimum Wiehler, *A. antirrhina (DC.) C.V.Morton, A. (Poeppig) G.Bentham ex Walpers, D. urticifol­ brevifolia C.V.Morton, *A. candida Lindley, *A. ium (Poeppig) G.Bentham ex Walpers, D. vex­ cettoana H.E.Moore, *A. dulcis C.V.Morton, *A. ans H.E.Moore, D. weberbaueri Fritsch, and D. erecta (Lamarck) H.P.Fuchs, *A. jimbriata Rose williamsii Rusby. ex C.V.Morton, *A. flava C.V.Morton, *A. gla­ Although badly in need of revision, Diastema, brata (Zuccarini) Fritsch, *A. grandiflora as a genus, is morphologically well defined; and (Schiede) DC., *A. heterophylla (C.RP.Martius) there is little doubt about its generic boundaries DC., *A. hintoniana A.Ramfrez-Roa & (but see below). Among its key characters are a L.E.Skog, *A. longiflora DC., *A. mexicana racemose flowering axis consisting of solitary (B.C.Seemann) G.Bentham & I.D.Hooker ex flowers in the axils of bracts on stems with (usu­ Fritsch, *A. misera Lindley, *A. nayaritensis ally) condensed internodes; a nectary consisting L.E.Skog, A. obscura C.V.Morton, *A. occiden­ of 5 long, finger-like glands; and a distinctive talis C.V.Morton, *A. patens G.Bentham, *A. bilabiate stigma. In addition, most (but not all) pedunculata G.Bentham, A. saxicola (Brande­ species have small white flowers with a single gee) C.V.Morton, A. skinneri Lindley, *A. war­ purple blotch on each lobe. All of these are char­ szewicziana (Regel) H.E.Moore, and *A. woodii acters not found in other taxa of Gloxinieae, in­ C. V.Morton. cluding Gloxinella, which is the probable sister­ Achimenes has undergone reorganization sev­ taxon of Diastema. eral times in the last 30 years (particularly Wieh­ Although collections usually can be assigned ler 1976, Ramfrez-Roa 1987). Molecular phy­ easily to this genus, assigning them to a species logenetic tools recently have been used to ex­ is more problematic. Forty-six names have been plore phylogenetic relationships and floral evo­ described in Diastema, but it is unclear how lution in the genus (Roalson et al. 2003). There many species should be recognized; several spe­ appear to be three or four major lineages of cies are known only from their type collections Achimenes (Roalson et al. 2003, E. Roalson et and should probably be synonymized under oth­ al. unpubl. data), and the genus may not be er species, whereas the circumscriptions of some monophyletic, with Solenophora possibly nested of the more common and widespread species within Achimenes (E. Roalson et al. unpubl. (e.g., D. racemiferum) may be overly broad. ROALSON ET AL.: GLOXINIEAE GENERIC REORGANIZATION 227

Diastema vexans H.E.Moore poses a particu­ Gloxinella lindeniana (Regel) E.H.Roalson & lar problem. Although agreeing with other spe­ 1.K.Boggan, comb. nov. Basionym: Tydaea cies of Diastema in several characters (nectary lindeniana Regel, Gartentlora 17: 257, pI. configuration, stigma type, fruit type, and a 589. 1868. Synonyms: Gloxinia lindeniana small white flower with a purple blotch on each (Regel) Fritsch, Oesterr. Bot. Zeit. 63: 66. lobe), D. vexans differs in its inflorescence 1913. Kohleria lindeniana (Regel) H.E. structure (1-4 flowered, usually bracteolate, Moore, Gentes Herb. 8: 380. 1954. TYPE: pair-flowered cymes in the axils of foliar leaves). Regel, s.n. (LE, not seen). Phylogenetic studies (Roalson et al. 2005) show D. vexans to belong to a clade including Koh­ Tydaea lindeniana Regel has been shuttled leria and Pearcea rather than the clade with all among several genera in its taxonomic history, other Diastema. In this case, the discrepancy be­ suggesting that it fits well into none of them. tween morphology and phylogeny is especially Placed in Gloxinia by Fritsch (1913), it was later striking, and D. vexans and its relationship to the transferred to Kohleria by Moore (1954), who rest of the genus should be further investigated. created the monotypic section Gloxinella to ac­ One possibility is that D. vexans represents an commodate it. Wiehler (1976) transferred the ancient hybridization event between early mem­ species back to Gloxinia and listed Kohleria bers of the Pearcea/Kohleria and Diastema sect. Gloxinella as a synonym of Gloxinia. The clades (or conversely, that the rest of Diastema species is here removed from both genera, and had its origin in a hybrid between these two Moore's sectional name raised to generic rank, clades). As Diastema is in need of extensive re­ because this species does not belong to either of vision, and we have not examined the type spec­ the clades containing the type species of these imen of D. vexans, for now we retain this spe­ two genera (Roalson et al. 2005, E. Roalson et cies as a dubious member of Diastema. al. unpubl. data). Fruit characters, in particular, set this species Eucodonia Hanstein, Linnaea 26: 200-20l. morphologically apart from Gloxinia, as defined 1853 [Apr 1854]. TYPE SPECIES: Eucodonia here. Parsimony analyses (Roalson et al. 2005) ehrenbergii Hanstein (= E. verticillata weakly or moderately support G. Zindeniana, sis­ (Martens & Galeotti) Wiehler). ter to Diastema, or Diastema + Gloxinia dod­ The genus includes * Eucodonia andrieuxii sonii (=Nomopyle dodsonii, see below). Bayes­ (DC.) Wiehler and *E. verticillata (Martens & ian analyses moderately support the G. linden i­ Galeotti) Wiehler. ana/Diastema relationship (E. Roalson et al. un­ Molecular results (Zimmer et al. 2002, Roal­ publ. data). While the exact placement of G. son et al. 2003, Smith et al. 2004, Roalson et ai. lindeniana is somewhat variable depending on 2005, E. Roalson et al. unpubl. data) support the the analysis conducted, the species clearly is not separation of this genus from Achimenes as dis­ closely related to Gloxinia perennis. We are here cussed by Wiehler (1976, 1983). recognizing this species in its own genus rather than in its likely sister group, Diastema, since Gloxinella (H.E.Moore) E.H.Roalson & 1.K. G. lindeniana does not share several apparent Boggan, gen. nov., stat. nov. Kohleria sect. synapomorphies for Diastema (Diastema-type Gloxinella H.E.Moore, Gentes Herb. 8: 382. bilabiate stigma, nectary of 5 elongate, finger­ 1954. TYPE SPECIES: Gloxinella lindeniana like glands, flowering stems racemose). The (Regel) E.H.Roalson & 1.K.Boggan. general vegetative and floral aspect of Gloxinel­ Plants weak-stemmed herbs with scaly rhi­ Za lindeniana also is unlike that of any Diaste­ zomes, often produced at the tips of long stringy ma. rhizomes; indumentum villous, lacking uncinate Gloxinia L'Heritier, in Aiton, Hort. Kew 2: 331. trichomes. Leaves opposite, equal, with 6-8 1789. TYPE SPECIES: Gloxinia maculata pairs of veins. Flowers epedunculate, ebracteo­ L'Heritier, nom. illeg. (=Martynia perennis late, solitary in the leafaxils; corolla lavender, L.), Gloxinia perennis (L.) Fritsch, in A. lobes subequal, entire; nectary annular, some­ Engler & Prantl, Nat. Pflanzenfam. 4(3b): times slightly 5-lobed; ovary inferior; stigma 174. 1894. broadly stomatomorphic to obscurely bilobed. Fruit an ovoid to elliptic fleshy capsule, dehisc­ Plants erect herbs with scaly rhizomes (rhi­ ing along the dorsal side and splitting the hy­ zomes absent in Gloxinia xanthophylla), nearly panthium to the base. Seeds numerous, minute, glabrous to pilose, Jacking uncinate trichomes. rhombic to broadly ellipsoid, almost as broad as Leaves opposite (rarely temate), with 5-9 (-12) long. pairs of veins. Flowers raceme-like flowering The genus includes *Gloxinella lindeniana stems with leaves reduced to opposite or alter­ (Regel) E.H.Roalson & 1.K.Boggan. nate bracts bearing solitary ebracteolate flowers; 228 SELBYANA Volume 25(2) 2005 corolla white, pink, purple, or brownish, lobes Wiehler (1976), see the genera Gloxinella, Glox­ subequal to unequal, entire to toothed or fimbri­ iniopsis, Mandirola, Nomopyle, Seemannia, ate; nectary absent or annular; ovary half-infe­ Sphaerorrhiza, and incertae sedis species. rior to inferior; stigma capitate to stomato­ Among these species, Gloxiniopsis racemosa is morphic. Fruit an ovoid to elliptical dry rostrate most similar to Gloxinia perennis, but differs in capsule, loculicidally dehiscent without splitting other key characters (most notably the fruit); and the hypanthium. Seeds numerous, minute, rhom­ our phylogenetic analyses place it well outside bic to ellipsoid. the clade containing G. perennis. The genus includes *Gloxinia erinoides (DC.) Seemannia, Mandirola, and Goyazia are E.H.Roalson & 1.K.Boggan, *G. perennis (L.) members of a clade including Gloxinia as de­ Fritsch, and *G. xanthophylla (Poeppig) fined by us, but we consider them to be mor­ E.H.Roalson & 1.K.Boggan. phologically distinct enough to merit recognition at the generic level. Although the GloxinialSee­ Gloxinia erinoides (DC.) E.H.Roalson & mannialGoyazialMandirola clade could be par­ 1.K.Boggan, comb. nov. Basionym: Achi­ titioned in several ways, with anywhere from menes erinoides DC., Prodr. 7: 536. 1839. one to six monophyletic genera, we find that the Synonym: Koellikeria erinoides (DC.) generic circumscriptions proposed here best re­ R.Mansfeld, Repert. Spec. Nov. Regni Veg. flect the phylogenetic relationships and pattern 38: 28. 1935. TYPE: Venezuela-Districto of morphological variation. Given the shared in­ Federal, Vargas, 1. 1630 (holotype, G-DC). florescence characteristics of Gloxinia perennis, Gloxinia xanthophylla (Poeppig) E.H.Roalson G. erinoides, and G. xanthophyllus, this seems & 1.K.Boggan, comb. nov. Basionym: Ges­ to be a cohesive generic unit. Similarly, the his­ neria xanthophylla Poeppig, in Poeppig & torically recognized Seemannia forms a mono­ Endlicher, Nov. Gen. Sp. PI. 3: 7. 1840. phyletic group and can be defined by several Synonym: Anodiscus xanthophyllus (Poep­ synapomorphies, including stringy aerial rhi­ pig) R.Mansfeld, Repert. Spec. Nov. Regni zomes, an elongated pointed stigma, and barrel­ Veg. 36: 124. 1934. TYPE: Peru-Chihu­ shaped trichomes in the corolla. The generic aneila, E.Poeppig, s.n. (holotype, W). boundaries of Mandirola and Goyazia are some­ what more problematic as the species and ge­ Wiehler (1976, 1983) favored a broad circum­ neric boundaries of the included species are scription of Gloxinia, primarily on the basis of quite difficult to discern. We here have moved hybridization studies. Both molecular and mor­ some Gloxinia species into Mandirola rather phological studies indicate that Gloxinia sensu than combining them with Goyazia, which Wiehler is polyphyletic and requires extensive would require recognizing the entire group as reorganization. Mandirola rather than Goyazia; and the Man­ Phylogenetic studies demonstrate that the type dirola species do not share the Goyazia syna­ species of Gloxinia, G. perennis, is most closely pomorphies of pericraspedodromous-patterned related to Anodiscus xanthophyllus and Koelli­ leaf veins and a coriaceous leaf texture. While keria erinoides (Zimmer et al. 2002, Roalson et the exact boundaries of these genera require fur­ al. 2005), and this relationship is reflected in ther study, we consider this organization to be their morphology. Although a relationship be­ the most reasonable, as inferred by the distri­ tween Anodiscus and Koellikeria previously had bution of morphological characters and phylo­ been suggested (Wiehler 1983), the association genetic relationships (Roalson et al. 2005, E. of these genera with Gloxinia perennis (or with Roalson et al. unpubl. data). any species of Gloxinia s.l.) is novel. The most E.H.Roalson & 1.K.Boggan, gen. striking resemblance between these three species Gloxiniopsis is the raceme-like flowering stem, with flowers nov. TYPE SPECIES: Gloxiniopsis racemosa solitary in the axils of strongly reduced bract­ (G.Bentham) E.H.Roalson & 1.K.Boggan. like leaves. Although similar arrangements are Inter generes tribus Gloxinieae in fascibus vascular­ found in some other taxa in tribe Gloxinieae (no­ ibus petiolorum profunde lunate dispositis differt; a tably the genera Diastema, Gloxiniopsis, and Diastema, Monopyle, et Phinaea in corolla in calyce Smithiantha), we consider this general resem­ obliquo differt; a Gloxinia, Goyazia, Mandirola, et blance to reflect convergence rather than com­ Seemannia fere omni in hypanthio dorsaliter secedenti et in costis capsulae non prominentibus differt. mon ancestry. Given the close phylogenetic re­ lationship and the distinctive morphological Plants erect herbs with scaly rhizomes. similarity, we here transfer Koellikeria and An­ Leaves opposite, equal, with 9-12 (-14) pairs of odiscus to a much restricted (but morphologi­ veins. Flowers raceme-like flowering stems with cally better-defined) Gloxinia. leaves reduced to opposite bracts bearing soli­ For other species included in Gloxinia by tary ebracteolate flowers, white, campanulate, ROALSON ET AL.: GLOXINIEAE GENERIC REORGANIZATION 229 lobes subequal, lower lobes toothed; ovary in­ The genus includes Heppiella repens Han­ ferior; stigma stomatomorphic; nectary annular stein, *H. ulmifolia (Kunth) Hanstein, H. verti­ and strongly reduced or possibly absent. Fruit a cillata (Cavanilles) Cuatrecasas, and *H. viscida subglobose fleshy capsule, dehiscing along the (Lindley & J.Paxton) Fritsch. dorsal surface and splitting the hypanthium to Heppiella includes four species as circum­ the base. Seeds numerous, minute, almost as scribed by Kvist (1990), but the phylogenetic broad as long. position of H. repens and H. verticillata have The genus includes *Gloxiniopsis racemosa yet to be tested; and these species need to be (G.Bentham) E.H.Roalson & J.K.Boggan. included in phylogenetic analyses to verify the circumscription of Heppiella. Molecular analy­ Gloxiniopsis racemosa (G.Bentham) E.H. ses confirm the placement of Heppiella in tribe Roalson & J.K.Boggan, comb. nov. Basion­ Gloxinieae but do not suggest a close relation­ ym: Monopyle racemosa G.Bentham, Hook­ ship to any other genus (Roalson et al. 2005, E. er's Icon. PI. 12: 87. 1876; Bot. Mag. 102: Roalson et al. unpubl. data). pI. 6233. 1876. Synonym: Gloxinia race­ mosa (G.Bentham) Wiehler, Selbyana 1(4): Kohleria Regel, Index Sem. Turic. [4]. 1847, 387. 1976. TYPE: cultivated, collector un­ non Regel 1851. TYPE SPECIES: Kohleria known (holotype, K). hirsuta (Kunth) Regel, Flora 31: 250. 1848. Gloxiniopsis racemosa, previously Gloxinia The genus includes * Kohleria affinis (Fritsch) racemosa, does not appear to be closely related E.H.Roalson & J.K.Boggan, *K. allenii to any species yet sampled in phylogenetic anal­ P.C.Standley & L.O.Williams, *K. amabilis yses. One analysis (maximum parsimony anal­ (J.E.Planchon & Linden) Fritsch var. amabilis, ysis of ITS + trnL-F + morphology) suggests, K. amabilis var. bogotensis (G.Nicholson) however, that it might be sister to a clade con­ L.P.Kvist & L.E.Skog, K. bella C.V.Morton, K. taining Diastema, Gloxinella, Monopyle, Nom­ diastemoides L.P.Kvist & L.E.Skog, *K. gran­ opyle, plus a portion of Phinaea (Roalson et al. difiora L.P.Kvist & L.E.Skog, *K. hirsuta 2005). As with several other species previously (Kunth) Regel var. hirsuta, K. hirsuta var. lon­ treated in Gloxinia, Gloxiniopsis racemosa is gipes (G.Bentham) L.P.Kvist & L.E.Skog, K. clearly not closely related to the type of Glox­ hondensis (Kunth) Hanstein, K. inaequalis inia, G. perennis (Roalson et al. 2005, E. Roal­ (G.Bentham) Wiehler var. inaequalis, K. inae­ son et al. unpubl. data), despite a superficial qualis var. lindenii (Han stein) L.P.Kvist & morphological resemblance to this species. The L.E.Skog, K. inaequalis var. ocellata (W.J. generic name Gloxiniopsis refers to this similar­ Hooker) L.P.Kvist & L.E.Skog, K. longicalyx ity. L.P.Kvist & L.E.Skog, K. maculata C.V.Morton, K. neglecta L.P.Kvist & L.E.Skog, *K. peruvi­ Goyazia Taubert, Bot. Jahrb. Syst. 21: 451; 11 ana Fritsch, *K. rugata (Scheidweiler) L.P.Kvist Feb 1896. TYPE SPECIES: Goyazia rupicola & L.E.Skog, K. spicata (Kunth) Oersted, K. Taubert. stuebeliana Fritsch, *K. tigridia (J.H.Ohlen­ The genus includes Goyazia petraea dorf±) E.H.Roalson & J.K.Boggan, *K. trianae (S.M.Phillips) Wiehler and *G. rupicola Taub­ (Regel) Hanstein, K. tubiflora (Cavanilles) Han­ ert. stein, K. villosa (Fritsch) Wiehler var. aniso­ phylla (Fritsch) L.P.Kvist & L.E.Skog, *K. vil­ Molecular analyses show that Goyazia is closely related to Mandirola (see below), the losa var. villosa, *K. warsewiczii (Regel) Han­ species of which have a similar distribution in stein, * K. sp. nov. c2446, and * K. sp. aff. villosa Brazil. Future studies might support combining c6152 (At least one, and possibly two, unde­ the two into a single genus under the older name scribed species: K. sp. nov. Clark 2446 and K. Mandirola. We here maintain the two groups as sp. aff. villosa Clark 6152). separate (but closely related) genera, as they dif­ Kohleria affinis (Fritsch) E.H.Roalson & fer in numerous morphological characters. J.K.Boggan, comb. nov. Basionym: Capa­ One species formerly included in Goyazia, G. nea affinis Fritsch, Bot. Jahrb. Syst. 50: 434. villosa (Gardner) R.Howard (basionym Tapina 1913. TYPE: Colombia-Antioquia, Triana, villosa Gardner) was later transferred in Glox­ J. 2538 (holotype, W; isotypes, FI, G, K, inia by Wiehler (1976) but also is misplaced in MANCH, P, US). that genus; it is here considered incertae sedis Kohleria tigridia (J.H.Ohlendorf±) E.H.Roalson and possibly related to Phinaea. & J.K.Boggan, comb. nov. Basionym: Heppiella Regel, Gartenflora 2: 353. Dec 1853. Gloxinia tigridia J.H.Ohlendorff, in Otto & TYPE SPECIES: Heppiella viscida (Lindley & Dietrich, Allg. Gartenz. 13: 376. 1845. Syn­ J.Paxton) Fritsch. onyms: Besleria grandiflora Kunth, in 230 SELBYANA Volume 25(2) 2005

Humboldt, Bonpland & Kunth, Nov. Gen. sule. Seeds numerous, minute, rhombic to ellip­ Sp. PI. 2: qto. ed. 401, fol. ed. 321. 1818; soid. Colombia, A. Humboldt & A. Bonpland s.n. The genus includes *Mandirola ichthyostoma (holotype, P; isotype, P). Capanea grandi­ (Gardner) B.C.Seemann ex Hanstein, *M. mul­ flora (Kunth) J.Decaisne ex J.E.Planchon, tiflora (Gardner) J.Decaisne, and M. rupestris Fl. Serres Jard. Eur. 5: pI. 499-500. See (Gardner) E.H.Roalson & J.K.Boggan. Hanstein 34: 291. 1865. TYPE: described Mandirola ichthyostoma (Gardner) B.C. from cultivation (orig. colI. Moritz 1126: Seemann ex Hanstein, in C.EP. Martius, Fl. Merida, Venezuela) (holotype not seen). Brasil. 8(1): 348. 1864. Basionym: Gloxinia Recent phylogenetic studies (Smith et al. ichthyostoma Gardner, Hooker's Icon. PI. 5: 2004, Roalson et al. 2005, E. Roalson et al. un­ pI. 472. 1842. publ. data) have strongly supported the position Mandirola multiflora (Gardner) J.Decaisne, of Capanea nested within Kohleria, a possibility Rev. Hort. 20 [ser. 3, 2]: 468. 1848. Achi­ first raised in the revision of Kohleria by Kvist menes multiflora Gardner, Hooker's Icon. and Skog (1992), and it is therefore submerged PI. 5: pI. 468. 1842. Synonym: Gloxinia into Kohleria here. Because of the existing com­ planalta Wieh1er. Achimenes hirsuta DC., bination Kohleria grandiflora L.P.Kvist & non A. hirsuta Lindley (=A. skinneri Lind­ L.E.Skog, Besleria grandiflora Kunth cannot be ley). Gloxinia hirsuta (DC.) Wieh1er. transferred to Kohleria; and thus the next oldest Mandirola rupestris (Gardner) E.H.Roalson & epithet, Gloxinia tigridia J.H.Ohlendorff, must J.K.Boggan, comb. nov. Basionym: Achi­ be used. Kohleria tigridia is a widespread and menes rupestris Gardner, Hooker's Icon. PI. extremely variable species whose circumscrip­ 5: pI. 480. 1842. Synonym: Gloxinia rupes­ tion should be examined; it is possible that one tris (Gardner) Wiehler, Selbyana 1(4): 387. or more taxa synonymized under this name 1976. TYPE: Brazil-G. Gardner 3874 (ho­ should be recognized (J.L. Clark pers. comm.). lotype, K). Capanea has been traditionally separated from Kohleria by having an epiphytic habit, cap­ Phylogenetic studies clearly support a clade itate stigmas, and capsules that split with four including Gloxinia ichthyostoma Gardner and apical valves (versus terrestrial habit, bilobed Gloxinia planalta Wieh1er and, presumably, stigmas, and capsules splitting with usually two Gloxinia rupestris (Gardner) Wiehler (Roalson apical valves in Kohleria). Other characters of et al. 2005, E. Roalson et al' unpubl. data). The the calyx and corolla shape and inflorescence oldest generic name for this group is Mandirola structure are extremely similar among Capanea J.Decaisne. While material of Gloxinia rupestris and Kohleria species (Kvist & Skog 1992). has not been available for phylogenetic studies, These similarities in combination with the phy­ we have placed this species in Mandirola based logenetic position of Capanea nested within on strong morphological similarities and close Kohleria have suggested that the two species of geographical proximity to the other species here Capanea are best treated as specialized epiphyt­ considered in Mandirola (Roalson et al. 2005, ic Kohleria species. E. Roalson et al. unpubl. data). This genus was originally created to accommodate the Brazilian Mandirola J.Decaisne, Rev. Hort. 20 (ser 3. 2): species Achimenes multiflora Gardner, and in­ 468. 15 Dec 1848. TYPE SPECIES: Mandirola deed the species included here in Mandirola are multiflora (Gardner) J.Decaisne. Synonyms: extremely similar to Achimenes morphological­ Achimenes subg. Mandirola (J.Decais~e) ly. Phylogenetic studies, however, confirm that Hanstein, Linnaea 34: 343. 1865. Achi­ they are not closely related to that primarily menes sect. Mandirola (J.Decaisne) Central American genus. Chromosome numbers G.Bentham, in G. Bentham & Hooker, Gen. are unknown for this group; but based on phy­ PI. 2: 999. 1876. logenetic relationships, they are predicted to be n = 13 (as opposed to n = 11 in Achimenes). Plants erect herbs with scaly rhizomes. Despite the close resemblance of this group to Leaves subsessile to short-petiolate, opposite or Achimenes, their unknown chromosome num­ temate, equal to subequal, rarely unequal, with bers, and their absence from any hybridization 5-6 pairs of veins. Flowers in axillary bracteo­ studies, Wiehler (1976) transferred these species late cymes, often with a short peduncle (some­ from Achimenes to Gloxinia on the basis of hy­ times solitary); corolla pink, lavender, or purple, brids between- Gloxinia perennis and Gloxinia lobes subequal, usually distinctly toothed to fim­ gymnostoma, which he considered to represent briate; ovary half inferior; nectary annular; stig­ Achimenes section Mandirola. We, however, ma distinctly bilobed. Fruit a dry rostrate cap- have shown this species to belong to the See- ROALSON ET AL.: GLOXINIEAE GENERIC REORGANIZATION 231

mannia group (all of whose members hybridize Species of Monopyle can usually be distin­ easily with Gloxinia perennis) rather than the guished from other members of the same clade Mandirola group. Although phylogenetic studies by their anisophyllous leaves. Another character show this group to belong in the same major that was found to be restricted to Monopyle was clade as the type species of Gloxinia, the species the presence of uncinate (hooked) trichomes on of this group are morphologically and biogeo­ the calyx and hypanthium (and frequently other graphically distinct from our recircumscribed parts of the plant as well). Such trichomes were Gloxinia. Therefore we here resurrect the genus not observed in any other taxa, and while they Mandirola for this group to simultaneously rec­ are apparently lacking in some Monopyle spp. ognize its sister group, Goyazia, as a similarly (J. Boggan unpubl. data), uncinate trichomes distinct genus. Species circumscriptions within may be a useful character in distinguishing this Mandirola are difficult and need to be further genus from its relatives. explored. We have here moved Gloxinia refiexa into Monopyle, as one of its synonyms was once Monopyle Moritz ex G.Bentham, in G. Bentham treated, because it clearly does not belong to & J.D. Hooker, Gen. PI. 2: 997. May 1876. Gloxinia as here circumscribed, and it appears TYPE SPECIES: Monopyle leucantha Moritz to have the greatest similarity to Monopyle. ex G.Bentham, Icon. PI. 12: 87. 1876 (=M. While it does not share the Monopyle characters subdimidiata (Klotzsch & Hanstein) of unequal leaves and uncinate trichomes, we R.Mansfeld). believe this is a more reasonable place to treat The genus includes Monopyle angustifolia the species, until more detailed studies of this Fritsch, M. divaricata Rusby, M. ecuadorensis species and Monopyle as a whole can be con­ C.V.Morton, *M. fiava L.E.Skog, M. grandifiora ducted. Gloxinia refiexa and Monopyle do share Wiehler, M. inaequalis C.V.Morton, M. iserni­ the combination of an absent nectary and dis­ ana Cuatrecasas, M. macrocarpa G.Bentham tinctly unequal and oblique leaf bases, a com­ var. costaricana W.B.Hemsley, M. macrocarpa bination exceedingly rare except in these taxa. var. isophylla G.Bentham, *M. macrocarpa var. A thorough revision of Monopyle is badly macrocarpa, M. maxonii C.V.Morton, M. mexiae needed. Rather than expand the circumscription C.V.Morton, M. panamensis C.V.Morton, M. of Monopyle (beyond moving Gloxinia refiexa paniculata G.Bentham, *M. puberula C.V.Morton, back into Monopyle) to include several taxa that M. sodiroana Fritsch, M. stenoloba C.V.Morton, are clearly related and share some characters M. subdimidiata (Klotzsch & Hanstein) with this genus but are otherwise morphologi­ R.Mansfeld, and M. subsessilis G.Bentham. cally heterogenous, we have retained a narrow circumscription to maintain the morphologically Monopyle reflex a (Rusby) E.H.Roalson & well-defined genera Diastema and Phinaea. De­ 1.K.Boggan, comb. nov. Basionym: Glox­ fining a monophyletic Monopyle necessitates inia refiexa Rusby, Mem. Torrey Bot. CI. 6: creating several small genera (Gloxinella, Glox­ 94. 1896. TYPE: Bolivia-La Paz, M. Bang iniopsis, and Nomopyle, based on Gloxinia lin­ 1745 (holotype, NY; isotypes, A, BM, C, E, deniana, G. racemosa, G. dodsonii, respective­ F, G, GH, K, M, MANCH, MO, NY, PH, ly). Phylogenetic analyses of this clade includ­ US, W, WU). Synonym: Monopyle divari­ ing Diastema, Gloxinella, Gloxiniopsis, Mono­ cata Rusby, Bull. N. Y. Bot. Gard. 8(28): pyle, Nomopyle, and Phinaea (in part) using 119. 1912. nrDNA ITS, cpDNA trnL-F, and morphological The phylogenetic analyses of Roalson et al. cladistic data sets result in similar inferences of (2005) have shown that one of the defining char­ relationships as the analyses of the Gloxinieae acters of Monopyle, the fruit a fleshy capsule as a whole. The suggestion is that morphological splitting along the entire length of its dorsal sur­ homoplasy in the large analyses are not con­ face including the inferior portion, is shared with founding assessment of relationships within this several species previously classified in Gloxinia clade of genera (E. Roalson unpubl. data). (G. lindeniana, G. racemosa, G. dodsonii); in­ Moussonia Regel, Index Sem. Turic. [4]. 1847. deed some of these taxa (e.g., G. racemosa) pre­ TYPE SPECIES: Moussonia deppeana (D.F.L. viously have been included in Monopyle. A very Schlechtendal & Chamisso) Hanstein. similar fruit also is shared with two taxa not in­ cluded in our analyses, Gloxinia refiexa and Ni­ The genus includes Moussonia ampla phaea peru viana. Apparently this fruit type is a L.E.Skog, *M. deppeana (D.F.L.Schlechtendal synapomorphy for this entire clade (rather than & Chamisso) Hanstein, *M. elegans J.Decaisne, a synapomorphy for Monopyle), but the precise M. fruticosa (Brandegee ) Wiehler, M. hirsutis­ relationship among these taxa requires further sima (C.V.Morton) Wiehler, M. rupicola study. (P.C.Standley & L.O.Williams) Wiehler, *M. 232 SELBYANA Volume 25(2) 2005

septentrionalis (D.L.Denham) Wiehler, M. ser­ Dodson & L. Thien 1173 (holotype, SEL; rulata (C.VMorton) Wiehler, M. skutchii isotypes, BH, K, UC, US, WIS). (C.VMorton & D.N.Gibson) Wiehler, M. stri­ Nomopyle peruviana (Wiehler) E.H.Roalson & gosa (C.v.Morton) Wiehler, M. triflora (Martens 1.K.Boggan, comb. nov. Basionym: Ni­ & Galeotti) Hanstein, and M. viminalis (Bran­ phaea peruviana Wiehler, Gesneriana I: 65. degee) Wiehler. fig. 18. 1995. TYPE: Peru-Huanuco, R.L. Molecular results (Zimmer et aI. 2002, Roal­ Dressler 4935 (holotype, SEL). son et aI. 2003, Smith et aI. 2004, Roalson et aI. 2005, E. Roalson et aI. unpubJ. data) support the Nomopyle dodsonii, previously Gloxinia dod­ separation of this genus from Kohleria as dis­ sonii, clearly is not closely related to Gloxinia cussed by Wiehler (1975, 1983). Species of perennis, and appears to have affinities to a Moussonia are unusual in tribe Gloxinieae, as clade containing Diastema, Gloxinella, Mono­ they do not produce scaly rhizomes; but molec­ pyle, and a portion of Phinaea, or possibly the ular and morphological analyses place this genus Heppiella lineage (Roalson et al. 2005, E. Roal­ in tribe Gloxinieae and suggest that the the taxa son et al. unpubI. data). The exact affinities of are secondarily arhizomatous (Roalson et aI. this species are not entirely clear and need to be 2005). further explored, but it is not a Gloxinia, as de­ fined here. Nomopyle, an anagram of Monopyle, Niphaea Lindley, Bot. Reg. 27: Misc. 80; Oct is used here to reflect the similarities in mor­ 1841. TYPE SPECIES: Niphaea oblonga Lind­ phology of these species to Monopyle. ley. Niphaea peruviana Wiehler, while not yet in­ The genus includes Niphaea mexican a cluded in any molecular phylogenetic analyses, C.VMorton and *N. oblonga Lindley. clearly shares several characteristics with Nom­ The circumscription of this genus and its re­ opyle dodsonii. Particularly, Niphaea peruviana lationship to the two elements of the polyphy­ has stomata aggregated in groups like Nomopyle letic genus Phinaea, will be discussed in a sep­ dodsonii, a characteristic not found in any other arate paper (J. Boggan et al. unpubI. data). Gloxinieae species. Niphaea peruviana also is consistent with the Nomopyle dodsonii character Nomopyle E.H.Roalson & 1.K.Boggan, gen. of solitary and ebracteate axillary flowers and nov. TYPE SPECIES: Nomopyle dodsonii similarly is lacking the Monopyle apomorphies (Wiehler) E.H.Roalson & 1.K.Boggan. of anisophylly and uncinate trichomes. We be­ A GJoxinieae fere omni in rhizomatibus non squa­ lieve these similarities warrant inclusion of Ni­ matiferis, fructo triplo vel quadruplo longiore quam phaea peruviana in Nomopyle and therefore latiore et in stomatibus aggregatis differt; a Diastema make the transfer here. in stigmate indiviso non didymo differt; a Gloxinia, Goyazia, Mandirola, et Seemannia in hypanthio dor­ Pearcea Regel, Gartenflora J6: 388; Dec 1867. saliter secedenti differt. TYPE SPECIES: Pearcea hypocyrtiflora (l.D.Hooker) Regel. Plants weak-stemmed erect to decumbent gla­ brescent herbs, lacking uncinate trichomes; scaly The genus includes *Pearcea abunda (Wieh­ rhizomes present (Nomopyle peruviana) or ap­ ler) L.P.Kvist & L.E.Skog, P. bella L.P.Kvist & parently absent (N. dodsonii). Leaves opposite, L.E.Skog, P. bilabiata L.P.Kvist & L.E.Skog, P. equal, with 5-8 pairs of veins, undersides with cordata L.P.Kvist & L.E.Skog, P. fuscicalyx stomata in indistinct groups. Flowers epedun­ L.P.Kvist & L.E.Skog, P. glabrata L.P.Kvist & culate, ebracteolate, solitary in the leafaxils; co­ L.E.Skog, P. gracilis L.P.Kvist & L.E.Skog, P. rolla campanulate to almost rotate, white to lav­ grandiflora L.P.Kvist & L.E.Skog, P. hispidis­ ender, lobes subequal, entire; ovary inferior; sima (Wiehler) L.P.Kvist & L.E.Skog, *P. hy­ stigma stomatomorphic; nectary absent or a re­ pocyrtiflora (J.D.Hooker) Regel, P. intermedia duced annular disc. Fruit a cylindric fleshy cap­ L.P.Kvist & L.E.Skog, P. purpurea (Poeppig) sule, dehiscing along the dorsal side and split­ L.P.Kvist & L.E.Skog, * P. reticulata (Fritsch) ting the hypanthium to the base. Seeds numer­ L.P.Kvist & L.E.Skog, P. rhodotricha (Cuatre­ ous, minute, subglobose. casas) L.P.Kvist & L.E.Skog, P. schimpfii The genus includes *Nomopyle dodsonii R.Mansfeld, P. sp. nov., P. sprucei (Britton) (Wiehler) E.H.RoaJson & 1.K.Boggan and N. pe­ L.P.Kvist & L.E.Skog var. parviflora (Rusby) ruviana (Wiehler) E.H.Roalson & 1.K.Boggan. L.P.Kvist & L.E.Skog, *P. sprucei var. sprucei, and P. strigosa L.P.Kvist & L.E.Skog. Nomopyle dodsonii (Wiehler) E.H.Roalson & Some previous authors have recognized the 1.K.Boggan, comb. nov. Basionym: Glox­ genus Parakohleria as separate from a mono­ inia dodsonii Wiehler, Selbyana 2(1): 80. pI. typic Pearcea (Wiehler 1978, 1983; Burtt & 24D. 1977. TYPE: Ecuador-Pichincha, C. Wiehler 1995). Phylogenetic studies have sup- ROALSON ET AL.: GLOXINIEAE GENERIC REORGANIZAnON 233 ported the position of the type species of Pear­ mann, Just's Bot. Jahresber. 26(1): 386. cea, P. hypocyrtiflora, as nested within the spe­ 1898. Synonyms: Fritschiantha nematan­ cies separated as Parakohleria (Roalson et al. thodes Kuntze; Gloxinia nematanthodes 2005, E. Roalson et al. unpubl. data). As the (Kuntze) Wiehler. recognition of Parakohleria would result in the Seemannia purpurascens Rusby, Mem. Torrey recognition of a paraphyletic genus, we recog­ Bot. Cl. 4: 237. 1895. Synonym: Gloxinia nize all of the species concerned as Pearcea purpurascens (Rusby) Wiehler. here, as suggested previously on the basis of morphological characters (Kvist & Skog 1996). Seemannia sylvatica (Kunth) Hanstein, Linnaea 29: 540, 587. 1859. Basionym: Gesneria Phinaea G.Bentham, in G. Bentham & J.D. sylvatica Kunth. Synonym: Gloxinia sylva­ Hooker, Gen. PI. 2: 991, 997. 1876. TYPE tica (Kunth) Wiehler. SPECIES: Phinaea albolineata (W.J.Hooker) We are here resurrecting the genus Seeman­ G.Bentham ex Hemsley. nia, as it forms a monophyletic and morpholog­ The genus includes * Phinaea albolineata ically well-defined group sister to an Anodiscusl (W.J.Hooker) G.Bentham ex Hemsley, "'P. mul­ Gloxinia perennislKoellikeria clade (Roalson et tiflora C.Y.Morton, and P. pulchella (Grisebach) al. 2005, E. Roalson et a!. unpub!. data). Be­ C. VMorton. cause of this close relationship, and because val­ Phylogenetic analyses show Phinaea to be id combinations for these species exist in the polyphyletic (Smith et al. 2004, Roalson et al. genus Gloxinia, an alternative classification 2005, E. Roalson et ai. unpubl. data), and mor­ would be to retain the Seemannia species in a phological studies corroborate this. A separate still-monophyletic but more heterogeneous cir­ paper (1. Boggan et ai. unpubl. data) will clarify cumscription of Gloxinia. We have chosen to re­ the circumscription of Phinaea (represented in store the generic status of Seemannia, as the spe­ our analyses by P. albolineata and P. mUltiflora) cies are distinctively different from the species and will include the description of a new genus we include in Gloxinia; and separating the two to accommodate Phinaea p.p. (represented in groups allows each genus to be defined more our analyses by P. divaricata and P. sp. nov. clearly (see discussion under Gloxinia). Seeman­ [96-336]). nia can be distinguished from other genera of Gloxinieae by the presence of barrel-shaped Seemannia Regel, Gartenflora 4: 183. 1855. multicellular trichomes in the corcola mouth and TYPE SPECIES: Seemannia ternifolia Regel an unusual pointed stigma. Seemannia species (=Seemannia sylvatica (Kunth) Hanstein). also have long whip-like aerial rhizomes in ad­ Synonyms: Fritschiantha Kuntze, Rev. dition to the typical Gloxinieae scaly rhizomes Gen. Pl. 3(2): 241. 1898. Fiehrigia Fritsch, and non-racemose inflorescences, both present Bot. Jahrb. Syst. 50: 397. 1913. in some other genera of the Gloxinieae, but rare­ Plants erect to decumbent herbs with scaly ly in this combination. rhizomes, often produced at the tips of long Smithiantha Kuntze, Rev. Gen. PI. 2: 977. stringy rhizomes. Leaves opposite, ternate, or 1891. Substitute name for Naegelia Regel whorled, equal, with 3-9 pairs of veins. Flowers 1847, not Naegelia Rabenhorst 1844, nor epedunculate, ebracteolate, usually solitary in Naegelia Zollinger & Moritzi 1845-1846, leafaxils (except Seemannia sylvatica frequent­ nor Naegelia Reinsch 1878. TYPE SPECIES: ly with 2-3+ flowers per axil); corolla tubular Smithiantha zehrina (Paxton) Kuntze. or inflated, often constricted at the mouth, red, orange, purple (rarely yellow), with barrel­ The genus includes *Smithiantha aurantiaca shaped multicellular trichomes at the mouth of Wiehler, *S. canarina Wiehler, S. cinnabarina the tube; lobes entire, subequal; nectary annular; (Linden) Kuntze, S. laui Wiehler, S. multiflora ovary half to almost completely inferior; stigma (Martens & Galeotti) Fritsch, and S. zebrina pointed. Fruit a dry rostrate capsule. Seeds nu­ (Paxton) Kuntze. merous, minute, ellipsoid. The geographically restricted and morpholog­ The genus includes * Seemannia gymnostoma ically well-defined Mexican endemic genus Smi­ (Grisebach) M.Toursarkissian, *S. nematantho­ thiantha is easily distinguished by its showy des (Kuntze) J.Schumann, *S. purpurascens flowers and racemose flowering stems with al­ Rusby, and *S. sylvatica (Kunth) Hanstein. ternate bracts, a combination not found in any other member of Gloxinieae in the same geo­ Seemannia gymnostoma (Grisebach) M.Tour- graphic area. sarkissian, Bol. Soc. Argent. Bot. 7(2): 135. Solenophora a.Bentham, PI. Hartw. 68; Mar 1958. Gloxinia gymnostoma Grisebach. 1840. TYPE SPECIES: Solenophora coccinea Seemannia nematanthodes (Kuntze) J.Schu- a.Bentham. 234 SELBYANA Volume 25(2) 2005

The genus includes Solenophora abietorum rolla mouth and lobes without stalked glandular P.C.Standley & Steyermark, *S. calycosa trichomes ...... Diastema (in part) J.D.Smith subsp. australis (C.V.Morton) Wei­ 6. Leaves usually in distinctly unequal pairs; bracts gend & Farther, S. calycosa subsp. calycosa, S. opposite; hypanthium and calyx usually with un­ cinate trichomes; fruit a fleshy capsule, dehiscing calycosa subsp. purpurascens Weigend & Farth­ along the dorsal side and splitting the hypanthium er, S. chiapensis D.N.Gibson, S. coccinea to the base ...... Monopyle (in part) G.Bentham, S. erubescens J.D.Smith, S. glom­ 6'. Leaves in equal or subequal pairs; bracts opposite erata Weigend & Farther, S. insignis (Martens or alternate; hypanthium and calyx never with un­ & Galeotti) Hanstein, S. maculata D.N.Gibson, cinate trichomes; fruit a dry or fleshy capsule .. S. obscura Hanstein, S. pirana C.V.Morton, S...... 7 purpusii Brandegee, S. schleehaufii Weigend & 7. Leaves with 3-5 pairs of leaf veins; bracts alter­ Forther, S. toucana D.L.Denham & D.N.Gibson, nate; nectary annular; fruit a dry capsule; n = 12; *S. tuerckheimiana J.D.Smith, S. tuxtlensis Ra­ plants of Mexico ...... Smithiantha mirez-Roa & Ibarra-Manriquez, and S. wilsonii 7'. Leaves with 5 or more pairs of leaf veins; bracts P.C.Standley. opposite or alternate; nectary annular or absent; n = 13 (where known); plants of South America, Solenophora recently has been revised, rec­ Central America, (West Indies by introduction), ognizing 18 taxa in 16 species (Weigend & not Mexico ...... 8 Forther 2002). Although Solenophora seems to 8. Leaves with 5-9 (rarely to 12) pairs of veins; nec­ be a well-defined genus morphologically, its tary annular or absent; fruit an ovoid, rostrate, dry possible entanglement with Achimenes needs to capsule, dehiscing dorsally and ventrally but not be further explored (see discussion under Achi­ rupturing the hypanthium; seeds longer than menes above). To date, only two of the 18 taxa broad; n = l3; plants of Central and South Amer­ have been included in phylogenetic analyses. ica (West Indies by introduction) . . .. Gloxinia More detailed phylogenetic analyses are neces­ 8'. Leaves with 9-12 or more pairs of veins; nectary sary to verify the circumscription of this genus. annular but strongly reduced (sometimes absent); fruit a subglobose fleshy capsule, dehiscing on the dorsal surface and splitting the hypanthium to A KEy TO THE GENERA OF GLOXINIEAE the base; seeds about as broad as long; plants of Colombia ...... Gloxiniopsis 1. Shrubby herbs to shrubs or small trees with soft 9. Small weak-stemmed herbs; flowers white, woody stems; scaly rhizomes never present; ova­ (sub)rotate with a very short tube ...... 10 ry half to fully inferior; plants of Mexico and 9'. Plant habit various; flowers various colors, more Central America ...... 2 or less zygomorphic with a distinct tube. .. 12 I' • Plant habit various, usually herbs to shrubby 10. Plants of Mexico and Guatemala; fruit an ovoid, herbs; scaly rhizomes usually present (absent in rostrate, dry capsule; filaments shorter than an­ some South American taxa); ovary inferior to al­ thers; n = 11; plants of Mexico and Guatemala most superior; plants of South America, Mexico, · ...... Niphaea Central America, and West Indies ...... 3 10'. Fruit a subglobose (rarely ovoid) dry or fleshy 2. Calyx lobes usually connate at least half their capsule; filaments longer than anthers; n = l3 or length (rarely free almost to ovary); flowers large 26; plants of Central America, South America, or and showy with a broad limb; ovary inferior; nec­ West Indies ...... 11 tary of one large (rarely 2-5) gland; fruit a glo­ 11. Fruit a fleshy capsule, held on erect pedicel above bose (sometimes fleshy) capsule, rupturing irreg- leaves, valves opening widely; plants of Central ularly; n = 10 ...... Solenophora America, South America, and West Indies .... 2'. Calyx lobes free almost to base; flowers narrow · ...... Phinaea s.s. and tubular with a small limb; ovary half inferior; 11'. Fruit a dry capsule, often held on pedicel curving nectary annular; fruit an ovoid, rostrate, dry bi­ below leaves, valves opening slightly; plants of valved capsule; n = 11 ...... Moussonia Central and South America . . . .. Phinaea p.p. 3. Flowering stems raceme-like with flowers solitary 12. Small herbs with wiry stems; leaves leathery, lat­ in the axils of small bract-like leaves ...... 4 eral veins reaching the margin and forming a mar- 3'. Flowering stems not raceme-like, flowers in pan­ ginal vein; plants of Brazil ...... Goyazia icles, axillary cymes or rarely solitary in leafaxils 12'. Plant habit various, leaves not leathery, lateral ...... 9 4. Bracts opposite; nectary consisting of 5 separate veins ending before reaching leaf margins .. 13 glands ...... 5 l3. Flowers tubular, red; anthers not coherent; fruit a 4'. Bracts opposite or alternate; nectary an annular dry capsule; plants of South America ...... ring or absent ...... 6 · ...... Heppiella 5. Robust strong-stemmed herbs; nectary glands 13'. Flowers various; anthers usually coherent; fruit a about as long as broad; stigma bilobed; corolla dry or fleshy capsule ...... 14 mouth and lobes usually with long-stalked glan- 14. Stigma distinctly bilobed ...... 15 dular trichomes ...... Kohleria (in part) 14'. Stigma various, but not distinctly bilobed .. 17 5'. Small weak-stemmed herbs; nectary glands fin­ 15. Nectary of separate glands (rarely united to some ger-like, longer than broad; stigma bilabiate; co- degree); corolla mouth andlor lobes usually with ROALSON ET AL.: GLOXINIEAE GENERIC REORGANIZATION 235

long-stalked glandular trichomes; fruit a dry or fruit cylindric; plants of Ecuador and Peru fleshy capsule ...... Kohleria (in part) ...... Nomopyle 15'. Nectary annular; corolla mouth and lobes without 24'. Leaves villous above, with stomata not in groups; stalked glandular trichomes; fruit a dry capsule nectary a slightly lobed annular disc; stigma ob­ · ...... 16 scurely bilobed; fruit ovoid to ellipsoid; plants of 16. Flowers usually in axillary cymes (rarely solitary Peru ...... Gloxinella in the leafaxils); corolla lobes usually distinctly toothed to fimbriate; plants of Brazil; chromo- SPECIES INCERTAE SEDIS some number unknown ...... Mandirola 16'. Flowers usually solitary in the 1eafaxils (some­ Gloxinia mieliezii Regel, Ind. Sem. Hort. Petrop. times in axillary cymes); corolla lobes usually en­ 1865: 64. 1865. tire; n = 11 or 22; plants of Central America and West Indies (elsewhere by introduction) .... This identity of the species is unclear, and the · ...... Achimenes (in part) original publication and type material have not 17. Nectary of 5 separate glands...... 18 been seen. 17'. Nectary annular or absent ...... 20 18. Small weak-stemmed herbs; flowers small, white Goyazia villosa (Gardner) R.Howard, J. Arnold with a purple spot on each lobe; nectary of 5 Arbor. 56(3): 367. 1975. Tapina villosa long, finger-like glands; stigma bilabiate; plants Gardner, Hooker's Icon. PI. 5: pI. 469. 1842. of Colombia ...... Diastema vexans Gloxinia villosa (Gardner) Wiehler, nom. il­ 18'. Plant habit and flowers various; nectary glands leg., non (Lindley) Martius; Selbyana 1(4): about as long as broad; stigma not dilapidate .. 387. 1976. · ...... 19 19. Plants terrestrial; rhizomes usually with widely Goyazia villosa (Gardner) R.Howard is here spaced scales; flowers usually red or yellow; considered incertae sedis because it is morpho­ plants of South America ...... Pearcea logically distinct from the Goyazia and Mandi­ 19'. Plants usually epiphytic; rhizomes usually absent rola groups. It is similarly out of place in Glox­ (if present, without fleshy scales); flowers usually inia. A relationship to Phinaea seems likely, but green or purple; plants of Central and South until the type can be examined, no transfer will America ...... Kohleria (Capanea group) be made. 20. Plants producing long stringy rhizomes in addi­ tion to scaly rhizomes; corolla with barrel-shaped multicellular trichomes at mouth; stigma pointed; GENERA EXCLUDED FROM THE fruit a dry capsule ...... Seemannia GLOXINIEAE 20'. Plants not usually producing long stringy rhi­ zomes; corolla without barrel-shaped trichomes; The genus Bellonia, variously placed in the stigma stomatomorphic or capitate, not pointed; Gloxinieae (Wiehler 1983, Burtt & Wiehler fruit a dry or fleshy capsule ...... 21 1995) or its own tribe (Fritsch 1893-1894), is 21. Fruit a dry capsule, not splitting the hypanthium most closely related to members of tribe Ges­ at deshicence; seeds usually longer than broad; n nerieae (Roalson et ai. 2005, E. Roalson et ai. = 11, 12, or 22; plants primarily of Mexico and unpubi. data). While Bellonia is not a close Central America ...... 22 morphological match to other genera of the 21'. Fruit a fleshy capsule, dehiscing along the dorsal surface, splitting hypanthium to base; seeds usu­ Gesnerieae, neither is it very similar to any of ally about as broad as long; n = 13; plants pri- the genera of the Gloxinieae, other than some marily of South America ...... 23 superficial floral resemblance with Niphaea 22. Stems and leaves densely lanate-villous; n = 12; and Phinaea (Xu & Skog 1990). The primary plants of Mexico ...... Eucodonia morphological characters linking it to Gesner­ 22'. Stems and leaves without woolly indumentum; n ieae are the absence of scaly rhizomes and = 11 or 22; plants of Mexico and Central Amer­ habit as a woody shrub. Bellonia also shares a ica (elsewhere probably by introduction) ... biogeographic distribution with the other gen­ · ...... Achimenes (in part) era of the Gesnerieae, as all but two or three 23. Leaves usually in distinctly anisophyllous pairs; species of this tribe are restricted to the Carib­ flowers in axillary, often pedunculate, cymes bean region. Given the molecular phylogenetic (sometimes in panicles, rarely solitary in the leaf data and the shared biogeographic distribu­ axils); hypanthium and calyx usually with unci­ tion, we consider Bellonia best treated as a nate trichomes; plants of Central and South America ...... Monopyle (in part) member of the tribe Gesnerieae despite its 23'. Leaves in (sub)equal pairs; flowers solitary (rare­ lacking several key apomorphies of other ly 2 or more) in the leafaxils; uncinate trichomes members of tribe Gesnerieae (e.g., fully infe­ absent; plants of South America ...... 24 rior ovary, spiral phyllotaxy, and chromosome 24. Leaves nearly glabrous above, undersides with number of n = 14). stomata in indistinct groups; nectary absent or a Detailed studies of Sinningia and relatives, reduced annular disc; stigma stomatomorphic; included in Gloxinieae by Wiehler (1983), are 236 SELBYANA Volume 25(2) 2005 underway by others (Perret et aI. 2001, Perret comb. nov. Basionym: Gloxinia sarmenti­ et aI. 2003). It is clear at this time from phy­ ana Gardner ex W.l.Hooker, Icon. PI. 4: pI. logenetic studies that these genera deserve to 378. 1841. TYPE: Brazil-G. Gardner 2226 be recognized as a tribe separate from the (holotype, K; isotypes, BM, CGE, K, L, P, Gloxinieae (Zimmer et aI. 2002, Mayer et aI. W). Synonyms: Gloxinia attenuata Hanst., 2003, Perret et aI. 2003, Weber 2004, Roalson Linnaea 27: 716. 1856. Gloxinia stolonifera et aI. 2005). Fritsch, Bot. lahrb. Syst. 37: 493. 1900. The genus Lembocarpus Leeuwenberg was classified in tribe Gloxinieae by Wiehler (1983), Recently the tribal affinity of Gloxinia sar­ and this placement was supported by Smith mentiana has been questioned (Zimmer et al. (2001). Although this paper does not address ge­ 2002). This species has been traditionally treated neric concepts in the Episcieae, it is clear at this in the genus Gloxinia (Hoehne 1964, Wiehler time (Roalson et aI. 2005) that Lembocarpus be­ 1976); but based on molecular phylogenetic longs in the Episcieae, as suggested by several studies (Zimmer et al. 2002, Roalson et al. authors (Beaufort-Murphy 1983, Boggan 1991, 2005), it clearly is not related to the genus Glox­ Feuillet & Skog 2003, Smith et al. 2004, 1.L. inia, nor does it even belong in tribe Gloxinieae. Clark unpubl. data). In its morphology, this species is consistent with members of Gloxinieae in some respects (rhi­ A NEw TRIBE OF GESNERIOIDEAE zomatous habit, semi-inferior ovary, annular nectary, and distinctly rostrate dry capsular Sphaerorrhizeae E.H.Roalson & 1.K.Boggan, fruit). In one key character, however, it differs; tribus nov. TYPE SPECIES: Sphaerorrhiza rather than producing rhizomes with fleshy sarmentiana (Gardner ex W.l.Hooker) scales, it produces rhizomes with small tuber­ E.H.Roalson & 1.K.Boggan. like swellings. In this respect, it more closely A Gloxinieae Fritsch in lobis calyce in alabastro val­ resembles the genus Sinningia (tribe Sinnin­ vatis, rhizomatibus tuberiferis differt. gieae), and several of the characters it shares with tribe Gloxinieae also are consistent with Sphaerorrhiza E.H.Roalson & 1.K.Boggan, tribe Sinningieae (e.g., semi-inferior ovary and gen. nov. TYPE SPECIES: Sphaerorrhiza rostrate dry caps1.dar fruit). sarmentiana (Gardner ex W.l.Hooker) From a phylogFnetic perspective, this species E.H.Roalson & 1.K.Boggan. does not share strong affinity with any other taxa A Gloxinia l'Heritier in lobis calyce in alabastro val­ sampled to date (Zimmer et al. 2002, Roalson et vatis, rhizomatibus tuberifero differt. al. 2005), although it weakly groups with tribe Plants erect to decumbent glabrescent herbs Sinningieae in some phylogenetic analyses producing rhizomes with tuber-like swellings, (Zimmer et al. 2002). Gloxinia burchellii ap­ often breaking apart with each propagule giving pears to share several characters not found in the rise to a new plant. Leaves short-petiolate to rest of Gloxinia s.l., and therefore tentatively is subsessile, opposite, equal, with 3-7 pairs of moved to Sphaerorrhiza, as S. burchellii, with veins. Flowers calyx lobes valvate and sealed in S. sarmentiana. Generally, this new genus seems bud; corolla broadly tubular, white, lavender, or to be a distinct lineage with phylogenetic prox­ purple; nectary annular; ovary half to almost imity to tribes Sinningieae and Episcieae (Zim­ completely inferior. Fruit a dry rostrate capsule. mer et al. 2002, Roalson et al. 2005); and the Seeds elliptic, numerous, minute. production of small tubers on underground rhi­ The genus includes Sphaerorrhiza burchellii zomes, a character shared with some members (S.M.Phillips) E.H.Roalson & 1.K.Boggan and of tribe Sinningieae, may provide a clue to the *S. sarmentiana (Gardner ex W.l.Hooker) origin of tuberous habit in that tribe. For these E.H.Roalson & 1.K.Boggan. reasons, we have erected a new tribe to deal with the lack of phylogenetic affinity of this genus to Sphaerorrhiza burchellii (S .M.Phillips) any current classification units. E.H.Roalson & 1.K.Boggan, comb. nov. Notably, Hoehne (1964) treated Gloxinia at­ Basionym: Achimenes burchellii S.M.Phillips, tenuata and G. stolonifera as valid species; de­ Kew Bull. 24(1): 225. 1970. Synonym: termining whether the variation within Sphae­ Gloxinia burchellii (S.M.Phillips) Wiehler, rorrhiza sarmentiana represents a single vari­ Selbyana 1(4): 387. 1976. TYPE: Brazil­ able taxon or two or more valid species will re­ Goias, W. Burchell 8615 (holotype, K; iso­ quire further study. Sphaerorrhiza is here types, L, WAG). presented as the generic epithet to reference the Sphaerorrhiza sarmentiana (Gardner ex distinctive rhizomes with tuber-like swellings W.l.Hooker) E.H.Roalson & 1.K.Boggan, found in the type species, S. sarmentiana. ROALSON ET AL.: GLOXINIEAE GENERIC REORGANIZATION 237

A KEy TO THE TRIBES OF GESNERIOIDEAE ACKNOWLEDGMENTS WITH INFERIOR AND HALF-INFERIOR OVARIES We thank Harold Robinson for help with the Latin diagnoses, Dan Nicolson for orthographic assistance, John L. Clark for helpful discussions, 1. Plants producing rhizomes with small tuber-like and an anonymous reviewer for helpful com­ swellings, but never with large perennial tubers or ments on a previous verison of the manuscript. scaly rhizomes; nectary annular; plants of Brazil ...... Sphaerorrhizeae 1'. Plants not producing rhizomes with tuber-like LITERATURE CITED swellings (if tuber-producing rhizomes present, then in addition to a large perennial tuber); nectary Beaufort-Murphy, H.T. 1983. The seed surface mor­ absent, annular or consisting of individual glands; phology of the Gesneriaceae utilizing the scanning plants of Central America, South America and electron microscope and a new system for diag­ West Indies ...... 2 nosing seed morphology. Selbyana 6(1-4): 220- 2. Plants usually with large perennial tubers, never 422. with SCalthizomes; nectary present, usually con­ Boggan, J.K. "A morphological study and cladistic sisting of 1-5 glands (if lacking tubers or with an­ analysis of Sinningia and associated genera with nular nec y, then plants of southern Brazil) ... particular reference to Lembocarpus, Lietzia, Pal­ ...... Sinningieae iavana, and Vanhouttea (Gesneriaceae: Gloxi­ 2'. Plants ne er with tubers; nectary absent, annular, nieae)." Thesis, Cornell University, Ithaca, New or consis 'ng of individual glands ...... 3 York, 1991. 3. Woody SJbShrubS, shrubs, or small trees, never Burtt, B.L. and H. Wiehler. 1995. Classification of the with scal rhizomes; leaves usually alternate, rare­ family Gesneriaceae. Gesneriana 1: 1-4. ly opposi e; ovary usually inferior (rarely half in­ Feuillet, C. and L.E. Skog. 2003. Novae Gesneriaceae ferior); n ctary annular (absent in Bellonia); fruit Neotropicarum XI. New genera and species from a dry cap ule; plants of the West Indies; 2 or 3 rare the Guianas. Brittonia 54: 344-351. Fritsch, K. 1893-1894. Gesneriaceae. Pp. 133-185 species J' Colombia and Venezuela ...... in A. Engler and K. Prantl, eds. Die natiirlichen 3'. Pflanzenfamilien, Vol. 4(3b). Wilhelm Engel­ H~~bs: ~. ~l~'~~~d; ~h~b~ ~~ ~~~i 'tr!:~::~:l~; mann, Leipzig, Germany. producin scaly rhizomes (if lacking scaly rhi­ ---. 1913. Gesneriaceen-Studien II. Uber Tydaea zomes, en not plants of southern Brazil); leaves lindeniana Regel. Oesterr. Bot. Zeit. 63: 64-67. opposite rarely whorled); ovary half inferior to in­ Hoehne, F.C. 1964.0 genero Gloxinia no Brazil. Arq. ferior (r ely almost superior); nectary absent, an­ Bot. Estado Sao Paulo 3: 315-335. nular, or onsisting of individual glands; fruit a dry Kvist, L.p. 1990. Revision of Heppiella (Gesneri­ or fleshy capsule; plants primarily of Central and aceae). Syst. Bot. 15: 720-735. South A erica (if West Indies, then with scaly rhi- Kvist, L.P. and L.E. Skog. 1992. Revision of Kohleria zomes) ...... Gloxinieae (Gesneriaceae). Smithsonian Contr. Bot. 79: 1-83. ---. 1996. Revision of Pearcea (Gesneriaceae). Smithsonian Contr. Bot. 84: 1-47. CONCLUSIONS Mayer, v., M. Moller, M. Perret, and A. Weber. 2003. Phylogenetic position and generic differentiation The clas ification changes proposed in this of Epithemateae (Gesneriaceae) inferred from plastid DNA sequence data. Amer. J. Bot. 90: paper will ring generic circumscription more 321-329. into line w th our current knowledge of phylo­ Moore, H.E., Jr. 1954. Some cultivated Gesneriaceae genetic reI tionships in the Gesnerioideae sub­ and hybrids. Gentes Herb. 8: 375-403. family, and particularly the Gloxinieae tribe, of Perret, M., A. Chautems, R. Spichiger, M. Peixoto, and Gesneriace e. Several issues cannot be ad­ V. Savolainen. 2001. Nectar sugar composition in dressed her , including the proper generic affin­ relation to pollination syndrome in Sinningieae ity of so e poorly understood species (e.g., (Gesneriaceae). Ann. Bot. 87: 267-273. Perret, M., A. Chautems, R. Spichiger, G. Kite, and V. Goyazia vi losa). In addition, generic circum­ Savolainen. 2003. Systematics and evolution of scription 0 species traditionally placed in Phi­ tribe Sinningieae (Gesneriaceae): evidence from naea will be addressed elsewhere. We have phylogenetic analyses of six plastid DNA regions made som§ assumptions about generic place­ and nuclear ncpGS. Amer. J. Bot. 90: 445-460. ment of sp cies within genera. Many of the gen­ Ramlrez-Roa, M.A. "Revision de Achimenes (Gesner­ era of the loxinieae need to be revised, includ­ iaceae)." Thesis, Universidad Nacional Autonoma ing detaile infrageneric phylogenetic analyses, de Mexico, D.F., 1987. Roalson, E.H., L.E. Skog, and E.A. Zimmer. 2003. to have more confidence in generic circumscrip­ Phylogenetic relationships and the diversification tions. This is particularly needed for the Achi­ of floral form in Achimenes (Gesneriaceae). Syst. menes/Solenophora circumscription, Diastema, Bot. 28: 593-608. Mandirola, Monopyle, and Moussonia. Roalson, E.H., J.K. Boggan, L.E. Skog, and E.A. Zim- 238 SELBYANA Volume 25(2) 2005

mer. 2005. Untangling the Gloxinieae (Gesneri­ ceae including Avicenniaceae). Springer-Verlag, aceae). I. Phylogenetic patterns and generic Berlin & Heidelberg, Germany. boundaries inferred from nuclear, chloroplast, and Weigend, M. and H. Forther. 2002. A revision of the morphological cladistic data sets. Taxon 54(2): Central American genus Solenophora (Gesneri­ 389-410. aceae). Harvard Pap. Bot. 7: 37-78. Skog, L.E. and J.K. Boggan. 2005. World Checklist of Wiehler, H. 1975. The re-establishment of Moussonia Gesneriaceae. Washington, DC: Dept. of Botany, Regel (Gesneriaceae). Selbyana 1 (1): 22-31. Smithsonian Institution. http://persoon.si.edu/ ---. 1976. A report on the classification of Achi­ Gesneriaceae/Checklist. Access date: I March menes, Eucodonia, Gloxinia, and Anetanthus 2005. (Gesneriaceae). Selbyana 1(4): 374-404. Smith, J.F. 200l. The phylogenetic relationships of ---. 1978. Parakohleria, a new South American Lembocarpus and Goyazia (Gesneriaceae) based genus in the Gesneriaceae. Selbyana 5(1): 4-10. on ndhF sequences. Ann. Missouri Bot. Gard. 88: ---. 1983. A synopsis of the neotropical Gesneri­ 135-143. aceae. Selbyana 6(1-4): 1-219. Smith, J.E, S.B. Draper, L.c. Hileman, and D.A. Xu, Z. and L.E. Skog. 1990. A study of Bellonia L. Baum. 2004. A phylogenetic analysis within tribes (Gesneriaceae). Acta Sci. Nat. Univ. Sunyatseni, Gloxinieae and Gesnerieae (Gesnerioideae: Ges­ Supp!. 9(2): 95-107. neriaceae). Syst. Bot. 29: 947-958. Zimmer, E.A., E.H. Roalson, L.E. Skog, 1.K. Boggan, Weber, A. 2004. Gesneriaceae. Pp. 63-158 in K. Ku­ and A. Idnurm. 2002. Phylogenetic relationships bitzki and l.W. Kadereit, eds. The Families and in the Gesnerioideae (Gesneriaceae) based on Genera of Vascular Plants, Vol. 7. Flowering nrDNA ITS and cpDNA trnL-F and trnE-T spacer Plants, Dicotyledons: Lamiales (except Acantha- region sequences. Amer. J. Bot. 89: 296-311.