Volume 11 NOVON Number 1 2001 New Combinations and New Names in Some Brazilian Microlicieae (Melastomataceae), with Notes on the Delimitation of Lavoisiera, Microlicia, and Trembleya Frank Almeda Department of Botany, California Academy of Sciences, Golden Gate Park, San Francisco, California 94118-4599, U.S.A. falmeda@calacademy.org Angela B. Martins Departamento de Botanica, IB, Universidade Estadual de Campinas, Caixa Postal 6109, 13083-970 Campinas, Sao Paulo, Brazil. amartins@unicamp.br ABSTRACT. Consideration of generic limits in the largely Brazilian tribe Microlicieae has resulted in an emended circumscription of the genus Lavoisi- era to include those species with a capsule that dehisces longitudinally from the base to the apex, and an ovary that is always partly inferior with a persistent columella and laterally flattened lamel- liform placental intrusions. An ongoing re-evalua- tion of Lavoisiera necessitates the transfer of seven species to Microlicia and one species to Trembleya. Six new combinations are proposed (Microlicia cer- ifera, M. mucugensis, M. noblickii, M. ordinata, M. vernicosa, and Trembleya elegans), and two new names are provided (Microlicia giuliettiana and M. longipedicellata) for epithets already pre-empted in the genus to which they are being transferred. Ge- neric limits in Microlicia and Trembleya are also discussed, together with distributional notes and di- agnostic characters for the species here transferred to these genera. Key words: Brazil, Lavoisiera, Melastomata- ceae, Microlicia, Microlicieae, Trembleya. The Microlicieae, with over 250 species, is the largest tribe of capsular-fruited Melastomataceae with a distribution centered in Brazil. Previous classifications of the family have attributed between 11 and 15 genera to this tribe (Cogniaux, 1891; Renner, 1993), which has traditionally been de- fined by its terete capsules, unadorned ovary apex, prolonged anther connectives, rostrate anther the- cae, and oblong or reniform seeds with a predom- inantly foveolate testa. We are currently preparing a monograph of La- voisiera DC., a genus of the Microlicieae with some 76 validly published species that is essentially re- stricted to campo rupestre habitats in central Bra- zil. Campo rupestre is a species-rich formation of interdigitating vegetation types dictated by slope, aspect, and drainage that is dominated by quartzitic outcrops with sandy or gravelly soils that are nu- trient-poor (Giulietti et al., 1987; Giulietti & Pirani, 1997; Stannard, 1995). The imprecise circumscription of genera in the Microlicieae has long been recognized (Baillon, 1877; Baumgratz et al., 1996; Hooker, 1867), yet no studies have attempted to evaluate intergeneric relationships within the tribe. To better understand the relationships and systematic position of Lavo- isiera, we have conducted a comprehensive char- acter analysis of all genera attributed to the Mi- crolicieae in the past. Although our work is still in progress, our phylogenetic analyses using morpho- logical characters show that Lavoisiera is consis- tently part of a clade that includes Chaetostoma DC., Microlicia D. Don, Rhynchanthera DC., Sten- NOVON 11: 1-7. 2001. Novon odon Naudin, and Trembleya DC. The remaining genera that have traditionally been included in the Microlicieae (Bucquetia DC., Cambessedesia DC., Castratella Naudin, Eriocnema Naudin, and Lith- obium Bongard) by Cogniaux (1891) and Renner (1993) are more distantly related to Lavoisiera. As- signment of these discordant genera to other cap- sular-fruited tribes such as the Melastomeae and Sonerileae (including Bertolonieae) may be neces- sary when seed characters are better understood. Of the five other genera with which Lavoisiera forms a distinctive clade, it appears to be most closely related to Microlicia and Trembleya. Within this group of three genera, Lavoisiera is distinguished by a capsular fruit that always dehisces in acropetal fashion from the base to the apex. It has a 4- to 8- locular ovary that is always partly inferior (gener- ally � inferior but always at least � inferior), a persistent columella (the central axis around which the carpels are arranged), and placentae in each ovary locule that form laterally flattened lamelli- form (platelike) intrusions. The flowers of Lavoisiera are typically 5- to 8-merous, rarely 9-merous. Many of the features that we use to characterize Lavoisi- era were also enumerated by Naudin (1844). How- ever, Naudin (1844) noted that the ovary of some species of Lavoisiera is 3-locular, and Cogniaux (1883) described and illustrated the ovary as 2- locular in a few species. We have been unable to verify the presence of a 2- or 3-locular ovary in any species of Lavoisiera and strongly suspect these were erroneous observations. With the exception of Lavoisiera, which is unique in the family, all other capsular-fruited melastomes exhibit basipetal dehiscence from the apex to the base of the capsule. Previous students of the Me- lastomataceae have provided no insights on the bi- ological significance of basal capsule dehiscence. We have given much thought to whether this ex- traordinary mode of fruit dehiscence has any adap- tive significance. Lavoisiera differs from other cap- sular-fruited melastomes of campo rupestre habitats in having ovaries that are partly inferior. This par- tially inferior ovary creates a fruiting structure that is typically thick-walled for �V to � of its length. Thus, one possible advantage of basal capsule de- hiscence is that it may promote rapid evacuation of mature seeds from the capsule. Apical dehiscence of a partly inferior capsular fruit with a similarly thick wall, on the other hand, creates a sturdier poorly dehiscent structure that would hold water longer and serve as a reservoir for the growth of destructive fungi and bacteria. In Microlicia, capsule dehiscence is always lon- gitudinal from the apex to the base, the ovary is always superior, the columella is deciduous, and the placentae in each ovary locule are dorso-ven- trally compressed and sometimes subpeltate. The flowers are typically 5-merous, very occasionally 6- merous, and 8-merous in one species. The ovary in Microlicia is typically 3-locular, rarely 3- and 4- locular on the same individual, and 5-locular in a few species. All of the largely Brazilian genera of Microli- cieae were probably derived from ancestral stock that was morphologically similar in many respects to Trembleya. This genus has capsules that dehisce basipetally from the apex to the base, the ovary is always superior, the columella is deciduous, the placental intrusions in each ovary locule are sub- peltate, the ovary is 3- to 5-locular, and the flowers are typically 5-merous, although two species are 4- or 5-merous. Other plesiomorphic characters of Trembleya are its pedicellate to subsessile flowers borne in dichasia or reduced modifications of di- chasia and the bracts and bracteoles that subtend nodes and pedicels of the inflorescence, respec- tively. This is in contrast to the situation in Mi- crolicia where the pedicellate to subsessile flowers are solitary and never subtended by modified leaves. Lavoisiera is more complex in this regard. It has two species with congested dichasia and as- sociated bracts and bracteoles. A great majority of its 35 to 40 species, however, have sessile solitary flowers that are typically subtended by modified leaves (bracts). The parallelisms in character states exhibited by these three genera are not unexpected in closely related taxa that have adapted to similar environ- ments with high insolation, pronounced seasonal rainfall, and a landscape characterized by a mosaic of nutrient-poor soils and highly dissected topog- raphy. Lavoisiera, Microlicia, and Trembleya have some consistent diagnostic characters that facilitate generic separation. They also exhibit some modally distinctive, but not always mutually exclusive, dif- ferences in floral merosity and ovary locule number that are clearly homoplasious. This has made the placement of anomalous or seemingly intermediate species difficult or arbitrary at times. The recog- nition of Lavoisiera as a natural genus is clearly defensible. The relationship between Microlicia and Trembleya, however, is very close. At this point in our studies of the Microlicieae, we defer to the judgment of a recent monographer (Martins, 1997) and recognize Trembleya at the generic level pend- ing the results of molecular data. Using the char- acter combinations enumerated above, we have been able to confidently place all described species of Lavoisiera into one of the three genera discussed Volume 11, Number 1 2001 Almeda & Martins Brazilian Microlicieae here. This has been possible even in the few cases where mature fruiting material was lacking. Based on our revised generic circumscription, we herein propose the transfer of seven species of Lavoisiera to Microlicia and one to the genus Trembleya. Microlicia cerifera (Gardner) A. B. Martins & Al- meda, comb. nov. Basionym: Lavoisiera ceri- fera Gardner, Sertum Plantarum part 3, tab. 63, 1844. TYPE: Brazil. Minas Gerais: elevat- ed mountain tracts NW of Diamantina, July 1840, G. Gardner 4581 (holotype, BM; iso- types, B-destroyed, F-frag., photo negative F 16653, K, NY, US). Distribution and phenology. Known only from the type, which was collected in flower in July. In the protologue, Gardner emphasized that this species differed from other described species of Ia- voisiera by its 3-celled ovary, but he expressed no doubts about his generic placement of the taxon. The 5-merous solitary flowers, lack of bracteoles, and 3-locular ovary of this species clearly dictate placement in Microlicia. None of the type speci- mens have capsules that are sufficiently mature to determine the exact nature of capsule dehiscence, but we feel confident that they will prove to be apically dehiscent when good fruiting material comes to light. In addition to the above characters, M. cerifera is distinguished by its 3-nerved leaves that are sessile and semiamplexicaul. Microlicia giuliettiana A. B. Martins & Almeda, nom. nov. Basionym: Lavoisiera luetzelburgii Markgraf, Notizbl. Bot. Gart. Berlin-Dahlern 10: 47. 1927. TYPE: Brazil. Bahia: 1914, P Luetzelburg 286 (holotype, M). Distribution and phenology. Endemic to the Chapada Diamantina in Bahia, Brazil, from the vi- cinity of Lencois and Mucuge west to Piata and south to Pico das Almas and the vicinity of Rio de Contas where it is locally common in campo ru- pestre and grassy slopes at 700-1850 m. Flowering from January through August, with an apparent peak in March, and in November and December; fruiting in December, February, March, and July (the peak month) and probably intervening months. A new name, Microlicia giuliettiana, is provided for this species because the basionym is already pre-empted by M. luetzelburgii Markgraf (Markgraf, 1927). Markgrafs placement of this species in La- voisiera is puzzling because he described it as hav- ing solitary 5-merous flowers and a 3-locular ovary, characters traditionally associated with Microlicia. Our examination of material consistently shows that the capsules dehisce from the apex to the base, the ovary is superior, the columella is deciduous, and the placental intrusions are dorso-ventrally com- pressed. Thus, its placement in Microlicia is clearly warranted. Woodgyer (1995) also questioned the ge- neric placement of Lavoisiera luetzelburgii and sug- gested that it might be more appropriately accom- modated in Microlicia. The unusual petal color pattern in this species is one of its most distinctive features. The petals are magenta pink when expanded with an abaxial red band or stripe on one side of each petal. The red band is typically all that is visible on floral buds and superficially gives the appearance that the petals are dark red. Other diagnostic features of this species are its obovate to broadly elliptic leaves that are apically rounded and mucronate with crenate to serrate margins beset with gland- tipped hairs. We take pleasure in naming this species for our colleague, Ana Maria Giulietti, in recognition of her many contributions to botanical research, the train- ing of many young Brazilian botanists, and her long-time efforts to promote an understanding of the rich campo rupestre flora in the states of Minas Gerais and Bahia. Brazil. Kpre'n'sernltilv .sperirmnes exam ined. BRAZIL.. Bahia: Municipio de ILenoi,. estrada dle Lencois BR 242. 5 km ao N dh ILenlois. 19 IDec. 1981. Cuaralho et al. 993 (CE- I:(c. I ): rgiio da Serra Sincor< . entre I uara e Mu- iisg'. 1M7 1t h. I)13. F'ris 20163. (1S): Muni-i pio de Pia- ta. Serra dot Aalho. proxinio a (;arimpo da (Cravada. 13,07'S. tI.l l 'W. 21 Aug. 1992. (Ganer 926 ((;AS, ill EFS); I kin SW of Mluciug on road from Cascavel near Fazenda P'aranguna. I "2.5' \. 13O()2'S. 6 Fleb. 1974. Har- lI 1078, (C1'KEC. N'. I S: l.wer NE slopes of the Pico ai-. Almnas. ca. 25 ki \X N\\ of the Vila do i de Contas. 11 57'\\. 13 33'S. 17 Fleb. 1977. H/arIe et al. 19503 (CE- I'(C. \N . I S): Serira do Ienqois. lower slopes of Morro do IPai Inmicio c. 11.5 km NW of I e-nquis just N of the main Seabra-ltnahraba road. 41�28'W. 12�27'S, 21 May 1980, HaIrle' et al. 22243 (CEP( IC. ILEC. S): Municipio de Piata. estrada para Inobia ca. 31 km de Piata. 15 Feb. 1987. Hlarl< et al. 24278 (CAS)>:M icfpio Rio de Con- las. Mato (;Iross. 1(I \lMa 1983. HIatsclhbach 46513 (CE- PI C. MIC I. IS). Microlicia longipedicellata Almeda & A. B. Martins. nomn. nov. Basionym: Lavoisiera glu- tinosa Cogniaux, in Mart. Fl. Bras. 14(3): 145. 1883. TYPE: Brazil. Minas Gerais: in locis saxosis prope Tejuco [Diamantina], Dec. 1824, L. Riedel 1224 (holotype, LE not seen, photo negative F 16660; isotypes, C, M, MO, NY, US, W). Distribution and phenology. Known only from the type and two other collections made in 1937, Novon all of which come from the Diamantina plateau in Minas Gerais, Brazil. These collections were gath- ered in November and December. Collections from both months have flowers; one of the collections made in December has persisting old capsules. In transferring Lavoisiera glutinosa Cogniaux to Microlicia we are providing a new name, Microlicia longipedicellata, because Microlicia glutinosa Nau- din is already pre-empted (Naudin, 1845: 180). In the protologue of Lavoisiera glutinosa, Cogniaux stated that the capsule opens from the base. This and the purported 6-locular ovary were probably the characters that prompted him to assign it to Lavoisiera. Our observations indicate that the cap- sules actually dehisce from the apex to the base. This kind of capsule dehiscence together with the ebracteate flowers, dehiscent columella, dorso-ven- trally compressed placentae, and superior ovary leave no doubt that L. glutinosa should be placed in the genus Microlicia as we circumscribe it. Mi- crolicia longipedicellata also has an 8-merous flow- er and an ovary that is 5-locular, character states that are common in species of Lavoisiera but un- common in Microlicia. In the past, a tendency to emphasize floral merosity and ovary locule number, to the exclusion of other characters, has led to the placement of anomalous species like this one in Lavoisiera instead of Microlicia. The specific epithet, longipedicellata, calls atten- tion to the persistent elongate pedicels of this spe- cies that measure 5-10 mm long. Microlicia lon- gipedicellata is also distinctive in having quadrate upper cauline internodes, glutinous trinerved leaves that are glandular-punctate with a conspic- uous network of prominulous venules on the abaxial surface, and triangular-subulate calyx lobes. Representative specimens examined. BRAZIL. Minas Gerais: Sentinella, Diamantina, 8 Nov. 1937. Mello Bar- reto 9575 (BHMH); Rio (rande, Diamantina, 8 Dec. 1937, Mello Barreto 10074 (F). Microlicia mucugensis (Wurdack) Almeda & A. B. Martins, comb. nov. Basionym: Lavoisiera mucugensis Wurdack, Phytologia 64: 294. 1988. TYPE: Brazil. Bahia: Mucuge, Corrego Moreira, 22 Jan. 1984, G. Hatschbach 47502 (holotype, MBM; isotypes, C, CEPEC, HUEFS, US). Distribution and phenology. Endemic to the Chapada Diamantina in Bahia, Brazil, from Anda- rai and Mucuge south toward Jussiape in campo rupestre at 900-1200 m. Flowering specimens have been collected in January, July, and September; good fruiting material has been collected in May but specimens collected in January also have old fruits. In addition to its 5-merous flower and 5-locular ovary, Microlicia mucugensis is distinguished by its viscose-punctate leaves that are sessile, ovate-ob- long, and cordulate at the base, as well as its com- pletely yellow anther thecae. Wurdack's (1988) decision to place this species in Lavoisiera was evidently influenced by ovary loc- ule number as the character of importance for ge- neric placement in the Microlicieae. On the basis of other fixed characters such as the apically de- hiscent capsules, superior ovary, deciduous colu- mella, and subpeltate placental intrusions, none of which were considered diagnostic by Wurdack, as- signment of this species to Microlicia is consistent with our circumscription of these genera. In the protologue, Wurdack noted that M. mu- cugensis is most closely related to Lavoisiera glu- tinosa (in agreement with our transfer of the latter to Microlicia as M. longipedicellata). The latter spe- cies differs in its 6-merous flowers, ovate leaf blades that taper to the base, and well-developed pedicels (5-10 mm). Wurdack also emphasized the strong vegetative resemblance between M. mucu- gensis and M. hatschbachii Wurdack, noting that the latter species has a 3-locular ovary, basally acute leaves, and shorter deltoid calyx lobes. Representative specimens examined. BRAZ I . Bahia: entre km 5-15 road, Mucuge rodovia para Andarai, 15 Sep. 1985, Halschbach 48248 (C, US); Municipio de Mu- cuge 3 km ao S de Mucuge, na estrada para Jussiape, 13)00'S, 41�24'W, 26 July 1979, Mori et al. 12559 (US): Municipio de Mucuge, nov a rooi Mucug/Andarai en- tre os km 0 e 1) 1 9 May 1989. Silra et al. 2774 (UB). Microlicia noblickii (Wurdack) A. B. Martins & Almeda, comb. nov. Basionym: Lavoisiera nob- lickii Wurdack, Kew Bull. 50: 821. 1995. TYPE: Brazil. Bahia: Palmeiras, 19 Nov. 1983, L. R. Noblick & A. Pinto 2769 (holotype, HUEFS; isotypes, CAS, CEPEC, US). Distribution and phenology. Endemic to the Chapada Diamantina in Bahia, Brazil, where it has been collected on and near Morro do Pai Inacio in campo rupestre off the road between Lencois and Palmeiras at 1000-1200 m. Flowering specimens have been collected in October and November; fruiting material has been collected in June with some October collections in old fruit. This species, like M. mucugensis, has 5-merous flowers and a 5-locular ovary. It is transferred to Microlicia for the same reasons enumerated in the discussion following that species. In his 1988 pro- tologue of Lavoisiera mucugensis, Wurdack also Volume 11, Number 1 2001 Almeda & Martins Brazilian Microlicieae noted that these two species are closely related. We agree with his assessment of interspecific relation- ships although we disagree with his placement of both species in Lavoisiera. Microlicia noblickii dif- fers from M. mucugensis in having larger leaf blades (12-16 X 10-11 mm vs. 7-10 X 6-8 mm), different calyx lobes (deltoid vs. lanceolate), and a longer connective prolongation on the larger an- thers (7 mm vs. 3.8-4.2 mm). Microlicia noblickii also differs from M. mucu- gensis in its polysporangiate anthers. Microlicia mucugensis, like the majority of angiosperms and presumably most Melastomataceae, has tetraspor- angiate anthers. Baumgratz et al. (1996) first de- scribed polysporangiate (multilocular) anthers in the family for two species of Chaetostoma and 14 species of Microlicia. The anthers of species that are polysporangiate have both of their thecae di- vided into numerous small locules in a way that resembles the structure of a honeycomb. The bio- logical significance of polysporangiate anthers is unclear at present. Vibrational pollination by bees is common in the Melastomataceae, and we have observed buzzing bees visiting flowers of Microlicia and Chaetostoma in the field. In polysporangiate anthers, Baumgratz et al. (1996) speculated that compartmentalization of the thecae would reduce the chance that all pollen is removed by the first insect visitor to a flower. Thus staggered pollen pre- sentation would insure pollen availability to more than one visitor and ultimate deposition on more than one stigma. Two of the species treated here, M. noblickii and M. ordinata, can now be added to the list of con- geners with polysporangiate anthers. As Baumgratz et al. noted, the polysporangiate condition in the Microlicieae may ultimately prove to be of taxo- nomic utility, at the species level at least. Our sur- vey of anther morphology in all species that we assign to Lavoisiera reveals that they are all tetras- porangiate. This corroborates and extends the find- ings of Baumgratz et al. who encountered only te- trasporangiate anthers in the 12 species of Lavoisiera examined for their study. Representative specimens examined. BRAZIL. Bahia: Municipio Palmeiras, Morro o Pai Inacio-Plato Cruz. 5 Jan. 1997. Conceiiao 218 (SPF): Municipio Palmeiras. Pai Inacio, 41�28'17"W. 12�27'31"S, 21 Nov. 1994. Melo et al. 1197 (ALCB); E-facing slope just below TV transmis- sion tower N of BR 242, ca. 5 km W of paved road to I,encois. 9 Nov. 1988. Kral et al. 75606 (US): Municipio of Palmeiras. Pai Inacio, BR 242. W of I.en 'ois at kmi 232. 12 June 1981, Mori & Boom 14368 (US). Microlicia ordinata (Wurdack) Almeda & A. B. Martins, comb. nov. Basionym: Lavoisiera or- dinata Wurdack, Phytologia 29: 135. 1974. TYPE: Brazil. Goias: Chapada dos Veadeiros, rocky slopes, 15 km W of Veadeiros, elev. 1000 m, 12 Feb. 1966, H. S. Irwin et al. 12695 (holotype, US; isotypes, F, MO, NY, RB, S, W). Distribution and phenology. Endemic to the Chapada dos Veadeiros in Goias, Brazil, where it is locally common in wet campo (brejo), grassy seep- ing slopes, wet sandy or rocky soil, and rocky slopes (campo rupestre) at 1000-1600 m. Flower- ing collections have been made from February through April; fruiting material has been collected in July and October. Microlicia ordinata, like M. mucugensis and M. nohlickii, also has a 5-merous flower and a 5-loc- ular ovary. Again, we suspect that the latter char- acter figured prominently in Wurdack's (1974) de- cision to assign it to Lavoisiera instead of Microlicia. An examination of other diagnostic characters shows that it has a superior ovary, apical capsule dehiscence, a deciduous columella, and dorso-ventrally compressed placental intrusions. In Wurdack's (1959) initial attempt to place this spe- cies generically, he commented on its superficial resemblance to Microlicia macrophylla Naudin, which has a 3-locular ovary. In the protologue he compared Lavoisiera ordinata to L. bicolor Naudin. The latter species, which appears to be known only from the type, has a 6-merous flower and a 6-loc- ular ovary (Cogniaux, 1883). Wurdack also com- mented on the resemblance of L. ordinata to M. pilosissima Cogniaux, a rare species that also has a 3-locular ovary. Some of the salient characters of M. ordinata in- clude its 4-winged upper internodes, ovate-orbic- ular, glandular pubescent leaves (0.5-1.4 cm long), short calyx lobes (1-2.8 cm), and a conspicuously 5-lobulate ovary apex. The anthers of M. ordinata are polysporangiate like those described above for M. noblickii. The chromosome number of M. ordinata was re- ported as n = 12 under the genus Lavoisiera (Al- meda, 1997). A gametic number of 12 is known for several species of Chaetostoma, Lavoisiera, Microl- icia, and Trembleya and appears to be the base number in the Microlicieae (Almeda, Martins & Romero, unpublished). Representative specimens examined. BRAZIL. Goias: 20 km by road N of Alto Paraiso, 5 Mar. 1973, Anderson et al. 6381 (C. F. MO. NY, 1B. US): 15 km S of Veadeiros. road to Sao Joao de Alian;a. 19 Mar. 1969. Irwin et al. 24639 (CAS): Chapada dos Veadeiros. rodovia GO-118. 4 km N de Alto Paraiso, 10 Feb. 1994. Hatschbach & Silra Novon 60297 (ltEFS): Chapatda dos \Vadeiros. 10 Km N of Al\to Paraiso. 24 Jan. 1980. King & Alrrred 8275 (C AS. M. MO, IB. L S). Microlicia vernicosa (Barreto ex Pedersoli) A. B. Martins & Almeda, comb. nov. Basionym: La- voisiera vernicosa Barreto ex Pedersoli, Orea- des 7 (12/13): 25-28. 1979/1980. TYPE: Bra- zil. Minas Gerais: Serra do Cip6, 13 Aug. 1933, H. L. M. Barreto 325 (holotype, BHMH). Distribution and phenology. Known only from the type, which was collected in flower in August. The protologue notes that fruits of this species were not seen. We suspect that its placement in Lavoisiera was probably based on its 6-merous flow- ers. Although the mode of capsule dehiscence can- not be determined on the basis of material at hand, it is clear that this species has a 3-locular ovary, which is never found in Lavoisiera. Because the ovary of M. vernicosa is superior, the ovary placen- tae are dorso-ventrally compressed, and the flowers are not subtended by bracts, we infer that this spe- cies will exhibit all the diagnostic characters that we attribute to Microlicia when fruiting material comes to light. In the protologue, M. vernicosa is compared with L. glutinosa (here transferred to Microlicia as M. longipedicellata). Both of these species have gluti- nous leaves and flowers that are solitary but aggre- gated in uppermost leafy branches. In M. longipe- dicellata the flowers are 8-merous, the ovary is 5-locular, the leaves are entire (vs. serrulate in M. vernicosa), and the pedicels are 5-10 mm (vs. 1-2 mm) long. Trembleya elegans (Cogniaux) Almeda & A. B. Martins, comb. nov. Basionym: Lavoisiera ele- gans Cogniaux, in Mart. Fl. Bras. 14(3): 160- 161. 1883. TYPE: Brazil. Goias: A. F. Glazioin 3758 (holotype, BR; isotype, C). Distribution and phenology. Probably wide- spread from the Carrancas region of southern Minas Gerais, Brazil, northwest to Serra da Canastra and north to an unspecified locality in Goias state in campo rupestre and seasonally wet campo at 1000 m. Flowering material has been collected from Feb- ruary through May and in October and November; fruiting material has been gathered in September and October. In the protologue, the type is cited as being from Rio de Janeiro but the label on the holotype gives the locality as Goias. No recently collected material has been seen from that state but six collections were made in Minas Gerais from 1996 through 1998. Cogniaux (1883) described this species as having solitary 4- or rarely 5-merous flowers, and a 4-locular ovary that is � inferior. The type does have a 4-locular ovary but it is superior. All recent collections of T elegans have a superior 5-locular ovary, an apically dehiscent capsule, and 5-merous flowers borne in dichasia with bracts subtending the inflorescence nodes and bracteoles subtending floral pedicels. The ovary also has subpeltate pla- cental intrusions that are typical of Trembleya. Until 1996, T elegans was known only from the type and one other collection. This probably ac- counts for the fact that its discordant position in Lavoisiera was not discovered for over 100 years. Its apparent rarity is also suggested by the fact that it is not conspecific with any of the described spe- cies of Trembleya recently treated by Martins (1997). Representatire splcimens examined. BRAZIL. Goias: in 1891-4-9 wilihout a specific locality. Glaziou 25297 (Bi). Minas Gerais: Municipio de Carrancas. caminho para (,ra dla l PoIte. 22 May 1997. latlsumoto et al. 293 (CA:S); tu:nicipio dle Carrancas. Poco da Ponte. 13 Sep. 1997. latsumaroto, 122 (L iC): Municipio de Carrancas. PI'oo di P'onte. It 39'\, 21�28'S. 10 Oct. 1997. Matsu- mnto et (il. 187 (CAS): Municipio te Carrancas. Poo da 'Pount. 10 Nov. 1997. Alatsumoto et al. 503 (UEC): Mun- icipio de Carrancas, To'a da Ponte, 6 Feb. 1998. MAtsu- moto et al. 655 (UlEC): Parquc Nacional da Serra da Can- astra, Sano Ioq(Ie de Minas, 24 Mar. 1996, Nakajima & Komero 1729 (CtAS). Acknowledgments. Fieldwork for this study was funded by the National Geographic Society (NGS Grant 6173-98 to Almeda). We are grateful to the Fundacao de Amparo a Pesquisa do Estado de Sao Paulo (FAPESP) for a postdoctoral grant (No. 99/ 0694-3) to Martins that made this collaborative pro- ject possible during her extended stay at the Cali- fornia Academy of Sciences. Martins also thanks the Conselho Nacional de Desenvolvimento Cientffico e Tecnolrgico (CNPq Grant No. 301346/86-0) for sup- porting her taxonomic studies of Melastomataceae. We also thank the Instituto Brasileiro do Meio Am- biente e dos Recursos Naturais Renovaveis (IBAMA) for permission to collect in protected areas of Brazil, the Universidade Estadual de Campinas, and the California Academy of Sciences for logistical and facilities support, and the following individuals who helped with many aspects of two recent field expe- ditions to Brazil: Orbelia Robinson, Don Robinson, Rosana Romero, Paulo J. F. Guimaraes, Renato Gol- denberg, Renato Belinello, and Jimi N. Nakajima. We also thank the curators and staffs of the following herbaria for loans, gifts, or special assistance during study visits: ALCB, B, BHCB, BHMH, BM, BR, C, CAS, CEPEC, DS, ESA, E HUEFS, HUFU, K, M, Volume 11, Number 1 2001 Almeda & Martins Brazilian Microlicieae RB, S, SP, SPF, UB, UEC, Literature Cited Almeda, F. 1997. Chromosome numbers and their evolu- tionary significance in some neotropical and paleotrop- ical Melastomataceae. BioLlania Ed. Esp. N� 6: 167- 190. Baillon, H. E. 1877. Melastomacees. Hist. P1. 7: 1-65. 1,. Hachette et Cie, Paris, London & Leipzig. Baumgratz, J. F. A., M. L. D. R. Souza, E. M. Woodgyer & E. M. Nic Lughadha. 1996. Polysporangiate anthers: Described for the first time in Melastomataceae. Kew Bull. 51: 133-144. Cogniaux, A. 1883. Melastomaceae (Microlicieae). In: C. F. P. von Martius, Flora Brasiliensis 14(3): 1-204. . 1891. Melastomacees. In: A. & C. de Candolle (editors), Monographiae phanerogamarum 7: 1-1256. G. Masson, Paris. Giulietti, A. M. & J. R. Pirani. 1997. Espinhaco Range Region, Eastern Brazil. Pp. 397-404 in S. D. Davis. V. H. Heywood, O. Herrera-MacBryde, J. Villa-Lobos & A. C. Hamilton (editors), Centres of Plant Diversity: A Guide and Strategy for Their Conservation (Vol. 3. The Americas). The World Wide Fund for Nature (WWF) and IUCN-The World Conservation Union. IUCN Pub- lications Unit, Cambridge, U.K. -- , N. L. Menezes, J. R. Pirani, M. Meguro & G. L. Wanderley. 1987. Flora da Serra do Cip6, Minas Gerais: MBM, MICH, MO, NY, R, US, W. Caracterizacao e lista das especies. Bol. Bot. Univ. Sao Paulo 9: 1-151. Hooker, J. D. 1867. Melastomaceae. In: G. Bentham & J. D. Hooker, Genera Plantarum 1: 725-773. Markgraf, 1927. Melastomataceae (Microlicia). In: R. Pil- ger (editor). Plantae Luetzelburgianae Brasilienses VII. Notizbl. Bot. Gart. Berlin-Dahlem 10: 45-47. Martins, E. 1997. Revisao Taxon6mica do Genero Trem- bleya DC. (Melastomataceae). D.Sc. Thesis, Univ. Es- tadual de Campinas, Sao Paulo, Brazil. Naudin, C. V. 1844. Additions a la flore du Bresil Merid- ional. Ann. Sci. Nat. Bot., Ser. 3, 2: 148-156. - 1845. Additions a la flore du Bresil Meridional. Ann. Sci. Nat. Bot., Ser. 3, 3: 169-192. Renner, S. S. 1993. Phylogeny and classification of the Melastomataceae and Memecylaceae. Nordic J. Bot. 13: 519-540. Stannard, B. L. (editor). 1995. Flora of the Pico das Al- mas. Chapada Diamantina-Bahia, Brazil. Royal Botanic Gardens. Kew. 'Woodgyer, E. 1995. Melastomataceae (Lavoisiera). Pp. 441-445 in B. L. Stannard (editor), Flora of the Pico das Almas. Chapada Diamantina-Bahia, Brazil. Royal Botanic Gardens, Kew. Wurdack, J. J. 1959. The Machris Brazilian Expedition: Phanerogamae, Melastomataceae and Polygalaceae. Los Angeles County Mus. Contr. Sci. 28: 3-11. . 1974. Certamen Melastomataceis XXIII. Phyto- logia 29: 135-151. -- . 1988. Certamen Melastomataceis XXXVIII. Phy- tologia 64: 293-301.