Selbyana 11: 26-34 REVISION OF HILLIA SUBG. RAVNIA (RUBIACEAE: CINCHONIOIDEAE) CHARLOTTE M. TAYLOR Department of Biology, University of Puerto Rico, Rio Piedras, Puerto Rico 00931 ABsTRACT. The characteristics used previously to separate Ravnia Oersted from Hillia Jacq. can now be considered mistaken or inconsistent. The subgenus Ravnia (Oersted) c. M. Taylor is erected in Hillia for three species, H. triflora (Oersted) C. M. Taylor, H. allenii C. M. Taylor (nom. nov.) and H. longiftl­ amentosa (Steyerm.) C. M. Taylor, and a new variety, H. triflora var. pittieri (Standley) C. M. Taylor, from Central America. Oersted (1852) erected Ravnia for a distinctive ment was made by Standley, although he had epiphytic shrub he found during his 1846-1848 seen more specimens (Standley, 1921) and first explorations of Costa Rica. In addition to its described the fruits (Standley, 1938). Steyermark succulent leaves and stems, R. triflora Oersted (1952) suggested that Ravnia was closely allied has showy bright red flowers 3-7 cm long with to Coutarea Aublet, but presented only the a curiously inflated corolla tube, which is swollen exserted stamens ofR. longifilamentosa Steyerm. in the middle portion but constricted at either as evidence for this. Ravnia has not been men­ end (FIGURE la, b). These flowers are similar to tioned in subsequent discussions of subfamilial those of some Gesneriaceae, such as species of classification in the Rubiaceae, although both Besleria, and in general aspect plants of Ravnia Verdcourt (1958) and Bremekamp (1966) have are easily mistaken for gesneriads. Oersted did suggested that Hillia deserves it own tribe based not see them, but the fruits of Ravnia are cylin­ on its plumose seeds. drical septicidal capsules with numerous small Determination of the relationships of Ravnia winged seeds, each bearing a tuft of long brown within the Rubiaceae has been complicated by trichomes at one end (FIGURE 1t). Oersted named a poor understanding of the genus itself. Al­ the plant to honor the Danish naturalist Peter though additional species have been published Ravn (1783-1839), who worked for the Botan­ (Standley, 1921; Steyermark, 1952), most tax­ ical Museum in Copenhagen and later collected onomists who have turned their attention to Rav­ specimens in Denmark's Caribbean possessions nia have finally combined all of the published (now the U.S. Virgin Islands) (Oersted, 1852; names, either explicitly (Dwyer, 1980) or in prac­ Vegter, 1983). Several species of Ravnia have tice (Standley, 1938: 1368). The resulting "R. been described since, all from Costa Rica and triflora" has become a variable taxon with a "re­ Panama. markable" (Dwyer, 1980) range ofleaf size, leaf Relatively few flowering specimens and no shape, anisophylly, corolla form and color, and fruiting material of Ravnia were collected in the sizes of floral parts. succeeding 80 years. Consequently published de­ Ravnia received monographic attention only scriptions of Ravnia remained incomplete, and from Standley (1921: 114). He recognized only its relationships within the Rubiaceae obscure. R. triflora and a new species, R. pittieri Standley With his original description Oersted also pre­ from Costa Rica. Later, however (1938: 1368), sented a diagnosis that separated Ravnia from he suggested that his new species might be "only Hillia Jacq., which he thought it strongly resem­ a narrow-leaved variety of R. triflora." Steyer­ bled. Schumann (1891: 51) later suggested in his mark (1952) described two more species from monograph of the world's Rubiaceae that from Panama, R. panamensis Steyerm. and R. lon­ Oersted's description, R. triflora was probably gifilamentosa Steyerm., but did not survey the allied to either Hillia or Cosmibuena Ruiz & genus. In the most recent consideration of the Pavon. No elaboration of this suggested place- genus, Dwyer (1980) combined all of the species ---+ FIGURE 1. a, Hillia triflora var. pittieri, habit with flower; from Nee 9733 (F). b, H. triflora var. triflora, habit with flowers; from Gomez 2230 (F). c, stipule; from Burger & Gentry 8569 (F). d, leaf of H. triflora var. triflora, showing narrow form; from Taylor & Skotak 4708 (DUKE). e, two capsules of H. triflora var. triflora; from Croat 46855 (MO). f, seed of H. triflora var. triflora; from Taylor & Skotak 4708 (DUKE). g, flower of H. longifilamentosa; from Allen 4744 (F). h, flower of H. allenii; from Allen 3558 (F). a-e, g to same scale. 26 1989] TAYLOR: HILLIA REVISION 27 8 u If', b d 28 SELBYANA [Volume 11 that have been described since 1852 with R. tri­ capitate in Hillia, in contrast to bilobed in Rav­ flora. nia, but some species described in Hillia have This study is a revision of Ravnia throughout bilobed styles (Steyermark, 1974: 165-178). All its range. Newly presented here are the extension authors have agreed that the calyx of Ravnia is of the geographic range, northward to southern persistent. Wernham (1916) contrasted this with Mexico and south to Colombia; expanded de­ the deciduous calyx of Hillia. However, Standley scriptions of all taxa; the revival ofSteyermark's (1938) used the persistent calyx of Hillia to sep­ two Panamanian species; the recognition of R. arate this genus from Cosmibuena. Calyx per­ pittieri Standley as a variety (but not a consis­ sistence therefore does not appear to be a con­ tently narrow-leaved one) of Hillia triflora; and sistent characteristic of Hillia. the transfer of all members of the genus to a The corollas of Hillia are white or green, fun­ newly-erected subgenus of Hillia, with the ap­ nelform to salverform, and often sweetly scented. propriate new combinations. A list of the spec­ These flowers are apparently adapted to moths imens examined, with determinations, is ap­ or other night-flying pollinators. In contrast, pended. Ravnia has bright red to dull pink, apparently This study is based on examination of her­ odorless corollas with a funne1form or tubular barium specimens from A, DUKE, F, GH, MO, NY, shape, which are apparently adapted for polli­ and SEL. Plants of Hillia are fleshy and shrinkage nation by hummingbirds. The corollas of both of all parts except woody stems accompanies the Hillia and Ravnia have a similar funnelform preparation of dried specimens. Calculations shape, and can be separated only by color. (The based on the dimensions ofthe fresh plants noted unique, swollen corollas of R. triflora are here on the labels of some specimens place this regarded as a derived character within these three shrinkage at about 8-10%. Measurements of soft red-flowered species.) Corolla color, then, is the parts given here are taken from dried specimens, only characteristic that remains to separate Rav­ and should be adjusted accordingly for fresh ma­ nia from Hillia. terial. This distinction between two closely related groups does not seem to merit recognition at the THE GENERIC PLACEMENT OF RA VNIA generic level. Other genera of Rubiaceae contain species that are apparently adapted to pollination Steyermark's (1952) suggestion that Ravnia is by different agents, with the corresponding vari­ allied to Coutarea is supported only by the fun­ ation in corolla color (for example Crusea Cham. nelform shape of the corolla in R. panamensis & Schldl., Hamelia Jacq., Isertia Schreber, and and R. longifilamentosa, and the six-merous Faramea Aublet). This variation does not weak­ flowers of some members of both genera. Cou­ en their identity. Therefore, Ravnia is here com­ tarea is a genus of terrestrial shrubs with small bined with Hillia. However, the Ravnia group stipules and zygomorphic flowers, and further of species shares corolla color and a relatively differs from Ravnia and Hillia in its glabrous restricted geographic range; these data may be seeds. As Schumann suggested (1891), Cosmi­ interpreted to indicate that these species are most buena is similar vegetatively and florally to species closely related to each other. Thus Ravnia is re­ of Ravnia and Hillia, but likewise has glabrous tained at the subgeneric level. The appropriate seeds. Hillia is the only genus in the Rubiaceae new combinations are made below. other than Ravnia with plumose seeds. In ad­ dition, Hillia shares floral and vegetative fea­ MORPHOWGY AND GEOGRAPHIC tures, except flower color, with Ravnia. DISTRIBUTION OF HILLIA SUBG. RA VNIA Without mature flowers, Hillia is difficult to separate from Ravnia. Previous authors have at­ The three species of Hillia subg. Ravnia C. M. tempted to distinguish them by the aestivation Taylor are succulent, usually epiphytic shrubs of the corolla lobes (Standley, 1921, 1938); the with relatively few, straggling, limber branches form of the stipules (Schumann, 1891); the form that may extend to I m or more in length. The of the style (Wernham, 1916); the degree of calyx stems are terete with smooth brown bark. Al­ persistence (Wernham, 1916); and the corolla though the plants are usually epiphytic, they are color and form (Dwyer, 1980). Corolla aestiva­ occasionally encountered in terrestrial sites, usu­ tion is generally described as convolute in Hillia ally in deep leaf mold or rotting wood. These in contrast to imbricate in Ravnia. However, all may be originally epiphytic plants that persist specimens of Ravnia examined have convolute after falling to the ground (F. Putz, pers. comm.). corolla lobes. The stipules have been described These species are found in wet forest from about as bilobed in Ravnia, in contrast to entire in 100 to 2,400 m elevation, from southern Mexico Hillia, but only entire stipules were seen in the to northern Colombia (FIGURE 2). Hillia triflora specimens examined. The style is supposedly (Oersted) c. M. Taylor is common both in ma- 1989] TAYLOR: HILLIA REVISION 29 20· N 15 FiGURE 2. a, distribution of Hillia triflora var. triflora (circles) and H. allenii (square). b, distribution of H. triflora var. pittieri (circles) and H. longifilamentosa (triangles). ture forest and in remnant woodlots and pasture the stamens alternate and adnate to the lobes.