Tibouchina triprovincialis R.Goldenb. & F.S.Mey., 2025

Goldenberg, Renato, Bigio, Narcisio C., Mendes-Silva, Ingrid, Medeiros, Herison, Koch, Ana Kelly, Ziemmer, Juliana Klostermann & Meyer, Fabrício Schmitz, 2025, Tibouchina triprovincialis, a new species of Melastomataceae from the Southwestern Amazonian grasslands, Phytotaxa 706 (1), pp. 61-73 : 63-70

Tibouchina triprovincialis R.Goldenb. & F.S.Mey. , sp. nov. ( Figures 1–4 )

Type:— BRAZIL, Rondônia: Machadinho D’Oeste, Distrito de Tabajara, Parque Nacional dos Campos Amazônicos , campo graminoso na cabeceira do igarapé do Marmelo, 8 o 55’4”S, 61 o 55’2”W, elev. 90 m, 10 March 2025. R. Goldenberg, H. Medeiros, I. Mendes-Silva, M. Nascimento 3514 (holotype: RON00026121 ; isotypes FLOR, NY, RB, UPCB0090998 ) .

Diagnosis:— Tibouchina triprovincialis differs from both T. aspera and Tibouchina spruceana Cogniaux (1885: 376) by the smaller leaves, with shorter petioles, 0.6–1.5 mm long (vs. 3–10 mm in T. spruceana , 2–10 mm in T. aspera ) and blades 5–11 mm long (vs. 20–50 mm in T. spruceana , 20–80 in T. aspera ), the abaxial surface densely covered with broad scales that conceal the surfaces of veins and the actual surfaces between the veins (vs. moderately covered with slender scales or trichomes that do not conceal the surfaces beneath, on both the veins and the actual surfaces between the veins).

Description:—Small shrub 20–80 cm tall, erect, the slender branches usually only a few, cylindrical in both younger and older branches; younger branches densely covered with light-brown to whitish scales, these on the internodes 0.7–2.2 × 0.2–0.4 mm, narrowly triangular to lanceolate, appressed, apex narrowly acute, apiculate, margins either densely lacerate-ciliate along its whole length (cilia up to 0.2 mm long) on the shorter scales, or remotely serrate and short ciliate (cilia up to 0.1 mm long) on the longer ones, on the nodes bigger, up to 4 × 0.8 mm, slightly erect, apex narrowly acute, apiculate. Leaves opposite, but sometimes with axillary sessile to very short branches that result in leaves resembling fasciculate, perpendicular to the stem (but somewhat erect in the axillary fascicles); petiole 0.5–1.5 mm long, broad, applanate and adaxially slightly folded, adaxial surface apparently glabrous, abaxial surface with scales 0.5–2.2 × 0.2–0.8mm, triangular to narrowly triangular, appressed, and also two similar but much bigger scales at the apex of the petiole that overlap the leaf base surface, these 3–3.5 x 0.7–1 mm, both with the apex acute to narrowly acute, apiculate, margins remotely serrate and short ciliate, petiole margins with triangular to lanceolate scales 0.7–1 × 0-2–0,25 mm, and also bigger scales 1.5-2.6 × 0.15-0.2 mm, linear lanceolate, both with the apex triangular to narrowly triangular, margins remotely serrate and short ciliate, mixed with vermiform trichomes; blade coriaceous, 5–11 × 2–6 mm, oval to broadly oval or broadly elliptic, seldom oval or elliptic, plane but with the base usually slightly adaxially folded, apex acute to slightly acuminate, always tipped with an apical, robust, prickly seta 0.2–0.6 mm long, base round to obtuse, margins scale-ciliate, the scales strongly appressed, 0.4–1 mm long, broadly triangular and adaxially folded, apex subulate and apiculate, base grooved (where it contacts the surface), margins shortly ciliate where facing the adaxial surface of the blade, lacerate and long ciliate where facing the abaxial surface, acrodromous veins 3, basal, conspicuous, plus a clearly visible, confluent, incomplete vein (i.e., each vein going from the base till ½ to ²⁄₃ of the blade’s length), adaxial surface bullate, with lines of bulla-based scales, each line in the space between and following two of the veins, and also between the external-most veins and margins, the actual scales on the top of the bullae slightly appressed, 0.4–0.8 × 0.15–0.3 mm, triangular to narrowly triangular, apex subulate and apiculate, base with grooves that extend to the adjacent bullae, margins lacerate and shortly ciliate, and also smaller lateral scales (on each bulla), similar to the previous ones, 0.1–0.3 × 0.1–0.15 mm, and also mixed with a few vermiform trichomes, abaxial surface foveolate, densely covered with two types of scales, the ones on the veins strongly appressed, 0.8–2.4 × 0.4–0.7 mm, thick, lanceolate, apex apiculate, margins lacerate and shortly ciliate (cilia up to 0.1 mm long), the ones on the foveolae slightly erect, but with a curved apex, 0.5–1 × 0.3–0.6 mm, more delicate than the ones on the veins, triangular, apex narrowly acute, margins lacerate and ciliate (cilia up to 0.1 mm long), mixed with dense vermiform trichomes. Inflorescences with apical, sessile, 4–9-flowered glomerules, seldom with one or two additional triads on the axils of the first pair of leaves right below; bracts 5–7, surrounding each glomerule, similar to the leaves but shorter, 4–9 × 3–4 mm; bracteoles 2, whitish to whitish-brown, both fused for about ½ of their length around each flower, sessile, 5–6 × 2.5–3 mm, ovate, elliptic or obovate, apex acute or obtuse, but always apiculate, persistent in young fruits, adaxial surface glabrous with a few slender scales on the apex (0.3–0.5 mm long) and base (1–1.5 mm long), abaxial surface covered with two layers of whitish to hyaline, appressed scales, the biggest ones 2–4.5 × 0.3–0.5 mm, lanceolate to linear lanceolate, apex narrowly acute and seta-tipped, margins sparsely ciliate, the smaller ones 0.5–1 × 0.1–0.2 mm, linear-lanceolate to linear, apex narrowly acute, margins repand. Flowers 5–merous, sessile; hypanthium whitish-green, ca. 5 × 4 mm, campanulate, adaxial surface glabrous, the surface smooth but bearing the impressions of the ovary scales on the abaxial surface, these circular to elliptic, abaxial surface densely covered with whitish to hyaline, appressed scales 0.8–2.6 × 0.3–0.5 mm, lanceolate, apex acute and seta-tipped, margins repand or remotely ciliate, the upper 10 scales arranged in a crown around the sepals, alternating intersepalar and sepalar (these right outside de sepals), similar to the ones on the rest of the hypanthium but somewhat bigger, 2.2–3.3 × 0.5–0.8 mm; sepals whitish-red outside and bright red inside, 4.2–5 × 1.4–1.6 mm, narrowly triangular, apex acute and seta-tipped, adaxial surface glabrous, abaxial surface moderate to densely covered with fine, hyaline, appressed scales 0.1–0.4 mm long; petals purple, 6–9 × ca. 5 mm, obovate, apex truncate or abruptly and shortly acuminate, with a scalariform seta 0.7–0.9 mm long, positioned in the middle of the truncate apex or on the acumen, always pointing downwards, margin moderately ciliate, with whitish trichomes or very narrow scales 0.1–0.6 mm long, erect, eglandular; stamens 10, slightly dimorphic in size, completely glabrous, the filaments whitish but becoming light-purple at the apex, pedoconnective light purple, straight, with white ventral appendages, these consisting of two rounded lobes 0.3–0.4 mm long in both cycles, anthers light purple but with the very tip yellowish around the pore, corrugated, subulate, in an almost 90 o angle with the pedoconnective, straight to slightly curved, the pore apical but slightly ventrally inclined, antesepalous with filaments 5.5–8 mm long, pedoconnective 1.5–2.5 mm prolonged below the thecae, anthers 3.5–4.5 × ca. 0.5 mm, antepetalous with filaments 4–7 mm long, pedoconnective 1–1.8 mm prolonged below the thecae, anthers 3.5–4 × ca. 0.6 mm; ovary superior, 5-locular, 2–3 × 2–2.5 mm, slightly 5-costate and the apex 5-lobed, densely covered with scales, these clustered in on each one of the five lobes from the distalmost point to half way to two thirds down to the ovary base, the whitish, erect scales ranging from 0.3–0.6 mm long closest to the base, then triangular to narrowly triangular, apex acute to narrowly acute, apiculate, to 1–2 mm long at the very apex, then narrowly triangular to linear, apex narrowly acute, apiculate; style 7.5–9 mm, filiform, straight, glabrous, stigma 0.4–0.5 mm diam., truncate. Velatidia with persistent sepals, 10–11 × ca. 3 mm (with the sepals) or 4.5–5 × ca. 3 mm (without the sepals), the five carpels (each keeping the scales on the apex) detaching from the central axis of the fruit, and this with persistent, perpendicular placentae; seeds 0.9–1 × 0.5–0.55 mm, cochleate, testa tuberculate.

Paratype s:— BRAZIL. Amazonas: Manicoré, Transamazônica, 149 Km sudeste de Humaitá, 5 Km Sul de Bodocó, “Campos naturais isolados”, 8 o 36’S, 61 o 25’W, 24 October 1993, T.M. Sanaiotti 274 (INPA, US); Trilha do encontro dos igarapés Macaco e Branco, Parque Nacional dos Campos Amazônicos, cerrado hiperestacional, 8 o 25’37”S, 61 o 43’69”W, 8 May 2009, L.C.P. Lima, L.P. Magalhães, A.S. Toledo, A.A. Souza 634 (HUEFS); Novo Aripuanã, Parque Nacional dos Campos Amazônicos, Cerrado Hiperestacional, 8 o 28’37”S, 61 o 13’85”W, 2 April 2018, C.B. Anjos, K.S. Gonçalves, K.R.C. Paixão, A.L.P. Silveira, J.S. Laurentino, F. Bonadeu 7 (RON). Mato Grosso: Colniza, Campinarana, ponto 28, MIR 270/271, 9 o 06’55” S, 60 o 48’27”W, alt. 214 m, 16 December 2020, S.S. Silva, J.O. Ala Filho, E.A. Silveira 23 (UFMT, UPCB). Rondônia: Machadinho D’Oeste, Vila de Tabajara, Parque Nacional dos Campos Amazônicos, Linha 5, Módulo 4 do RAPELD para o estudo Ambiental do AHE Tabajara, Savana Parque, UTM 20 L 621.419/9.007.204, February 2024. T.R.C. Capistrano et al. 114 (RON); Machadinho D’Oeste, Distrito de Tabajara, Parque Nacional dos Campos Amazônicos, Igarapé Marmelo, Campo Aberto, elev. 100m, 8°55’1’’S, 61°55’1”W, 10 March 2025, H. Medeiros, R. Goldenberg, I. Mendes-Silva, M. Nascimento 6337 (FLOR, NY, RB, RON, UPCB); Machadinho D’Oeste, Distrito de Tabajara, Parque Nacional dos Campos Amazônicos, Igarapé Marmelo, Campo Aberto, elev. 100m, 8°55’1’’S, 61°55’1”W, 10 March 2025, I. Mendes-Silva, R. Goldenberg, H. Medeiros, M. Nascimento 1190 (FLOR, NY, RB, RON, UPCB).

Distribution and habitat:—The distribution of T. triprovincialis is apparently restricted to Amazonian Brazil, in three states: Amazonas, Mato Grosso, and Rondônia ( Figure 5 ). All collections from Amazonas and Rondônia come from the “Parque Nacional dos Campos Amazônicos” (from now on PNCA, see D’Amico 2016). The single gathering outside PNCA is from northwestern Mato Grosso, which in turn was collected in the “Estação Ecológica do Rio Roosevelt” (EERR), a conservation area managed by the state of Mato Grosso (see https://uc.socioambiental. org/arp/1346); the label of Silva et al. 23 does not mention this conservation area, but the coordinates allowed us to place it in the EERR. Except for the three gatherings made by our team, the descriptions of the vegetation at the collection sites vary significantly among the respective collectors: Sanaiotti 274 (Amazonas) describes the place as “campo mal drenado” (poorly drained grasslands) and “campos naturais isolados” (natural and isolated grasslands); Anjos et al. 7 (Amazonas) call it “cerrado hiperestacional” (hyperseasonal savanna); Capistrano et al. 114 (Amazonas) say “savana parque” (savanna grasslands lacking shrubs or trees); Goldenberg et al. 3514, Medeiros et al. 6337 and Mendes-Silva et al. 1190 (Rondônia) describe the vegetation as “campos graminosos” (grasslands with Poaceae ); Silva et al. 23 (Mato Grosso) say “campinarana” (vegetation with shrubs and herbs on white sand). All seem to converge into a concept that describes a vegetation with a thick herbaceous layer and very few shrubs or trees, growing on poorly drained soils ( Figure 6 ). As explained in the introduction, this vegetation better fits the definition of “hyperseasonal savannas”, i.e. savannic enclaves in the middle of the Amazonian rainforest. Tibouchina triprovincialis seems to follow the same distribution pattern as Mabea dalyana ( Bigio et al. 2022) . Although other “hyperseasonal savannas” enclaves occur in southwestern Amazonia, they do not appear to host either of the two species described for the PNCA. These occurrences reinforce the uniqueness of this Amazonian vegetation and underscore the importance of the conservation units established to protect it.

We don’t have information on population density of T. triprovincialis for the other collections, but the ones visited by us were very extensive and dense ( Figure 6 ); we found hundreds of individuals growing among the grasses, in all areas where these grasses occurred, even in different grassland patches separated by arboreal vegetation.

Phenology:—Collected with flowers and fruits from February to May, and then October and December.

Conservation status:— Tibouchina triprovincialis has an EOO of 2,686.670 km ² and an AOO of 24.000 km ². In Western/Southern Amazonia, where this species is found, the major threat affecting biodiversity is the destruction of natural areas with both forest and grasslands, and their replacement with large-scale plantations or pastures. The EOO values and the number of threat situations are quite close to the thresholds for its inclusion in a threat category.

Most collections of T. triprovincialis were made inside the PNCA, while one of them was collected in EERR (see acronyms above), which means that they are currently protected. Nevertheless, both PNCA and EERR have been recently challenged as effective conservation areas. Part of the PNCA may be flooded if the Brazilian government approves a dam in the Rio Machadinho for a hydroelectric power plant ( D’Amico 2016). All conservation areas in the state of Mato Grosso have been threatened by invasions, both illegal and through lawsuits, and mismanagement, leading to criminal fires, degradation and loss of biodiversity ( Dorval et al. 2012). Despite being inside EERR, the specimen from Mato Grosso was collected at the roadside, in a region that is well known for land conflicts and illegal gold mining.

There is no data on population decline for the application of other criteria, but the populations that were visited by the authors and described right above (see Goldenberg et al. 3514, Medeiros et al. 6337, and Mendes-Silva et al. 1190) were enormous, with hundreds of individuals ( Figure 6 ). In conclusion, we suggest that T. triprovincialis should be regarded as Least Concern (LC).

Etymology:—The epithet “ triprovincialis ” refers to the fact that this species has been collected in three Brazilian states: Amazonas, Mato Grosso and Rondônia (see Figure 5 ).

Notes:—According to the modern circumscription of Tibouchina (Guimarães et al. 2025) , the plants in the genus are covered with an indumentum of scales, may have either fused or free bracteoles, and the stamens are glabrous or with trichomes over the filaments and pedoconnective appendages. The same authors define Tibouchina with “lilac or purple anthers, and petal color not changing with age”. According to this circumscription, T. triprovincialis belongs to Tibouchina , and not to the following two other genera in tribe Melastomateae Bartling (1830: 329 ; see Veranso-Libalah et al. 2022) with lepidote/scaly trichomes. Andesanthus has nine species restricted to the Andes and should be recognized by the “yellow anthers and isomorphic stamens, or if stamens dimorphic then with the larger anthers yellow and smaller anthers pink or red; petal color commonly changing with age” (ipsis litteris from Guimarães et al. 2019). Loricalepis Brade (1938: 71) has two species occurring on white sand vegetation, one in the Upper Rio Negro basin in Brazil and Colombia, and the other in the Atlantic coast in Bahia, also in Brazil ( Goldenberg et al. 2020); both species have smaller and much sparser scales, and also “young stem nodes with a transversal ridge, thickcartilaginous, crenulate-serrulate and paleaceous- ciliate leaf margins, petals with glandulose-ciliate margins, stamens with pedoconnectives not at all or little prolonged” (fide Goldenberg et al. 2020).

Tibouchina triprovincialis fits the circumscription of the former Tibouchina sect. Tibouchina ( Cogniaux, 1885, 1891) due to the vegetative parts covered with scales, the fused pair of bracts, and stamens with glabrous filaments and pedoconnective appendages. The only two species belonging to this section were T. aspera and T. spruceana , and no one else has been added to this pair since the 19 th century. The distinction between these two species is not very clear, mostly because of the huge morphological variation found in both, but mostly in T. aspera , for which three varieties have been described (see Cogniaux 1885). The two species differ mainly by the broader, 5-nerved leaves of T. aspera , when compared to the narrower, 3-nerved leaves of T. spruceana . Their wide distributions, as noticed by Jantzen et al. (2022), overlap in Colombia, Venezuela and northern Brazil, but T. aspera also occurs in Central America ( Belize and Nicaragua), Guianas and Peru. Their distribution in Brazil also overlaps with that of T. triprovincialis , except for the absence of T. spruceana in Rondônia (Flora e Funga do Brasil 2025). As already shown in the diagnosis, T. triprovincialis differs from both due to the smaller leaves, and the plants covered with denser and stouter scales all over. It also has bullate leaves with broader, triangular scales on the top of each bulla, on the adaxial surface. Its flowers are also slightly smaller, with the hypanthia ca. 5 mm, sepals 4.2–5 mm, and anthers 3.5–4.5 mm long, while in T. aspera the hypanthia are 7–8 mm long, the sepals 5–5.5 mm long, and the anthers 6–8 mm long, while in T. spruceana the hypanthia are ca. 5–6 mm long, the sepals ca. 5 mm, and the anthers 5–7 mm.

Geographically close to T. triprovincialis , most species of Tibouchina occurring to the south of the Amazon River (i.e., the “southern group” proposed by Jantzen et al. 2022) belong to the former Tibouchina sect. Barbigerae , and all differ from T. triprovincialis by the non-fused bracteoles and stamens with the pedoconnective appendages with a tuft of trichomes. Most of these species also have bigger leaves and thyrsoid inflorescences, instead of the small leaves and glomerules found in the new species. We opted not to compare T. triprovincialis with the species in this group, since the species that are morphologically closely related to it belong to the group discussed below.

All the geographically more distant “northern species” ( Jantzen et al. 2022) occur in Amazonian Venezuela (Wurdack 1993), with one species, T. catharinae Pittier (1947: 18) , that also occurs in adjacent Colombia and Brazil (Amazonas). They all differ from T. triprovincialis by the non-fused bracteoles, but they share with it the dense and robust scales all over (except for T. kunhardtii ), glabrous stamens and the smaller, more delicate plants, with small leaves. The two species that are morphologically closer to T. triprovincialis are T. dissitiflora Wurdack in Maguire & Wurdack (1958: 101) and T. huberi Wurdack (1990: 318-319 ; this species was not sampled by Jantzen et al. 2022). Both have slightly bigger leaves with a plane, non-bullate and striolate adaxial surface, and slightly bigger flowers. The leaves in T. dissitiflora have petioles 2–4 mm long and blades 10–25 × 8–15 mm, while T. huberi has petioles 3–7 mm and blades 15–45 × 10–20 mm, and T. triprovincialis has petioles 0.6–1.5 mm and blades 5–11 × 2–6 mm. The flowers in T. dissitiflora have hypanthia 6–6.5 mm long, sepals 7–11 mm long, and anthers 6–7.5 mm long, while T. huberi has hypanthia ca. 7 mm long, sepals ca. 6.5 mm, and anthers 6.8–8 mm, and M. triprovincialis has hypanthia ca. 5 mm, sepals 4.2–5 mm, and anthers 3.5–4.5 mm long.

Another taxon among the “northern species” ( Jantzen et al. 2022) is Tibouchina duidae Gleason in Steyermark (1952: 428), that differs from T. triprovincialis by the narrow 3-nerved leaves, with the lateral veins concealed by the scales and not visible on the adaxial surface (vs. lateral veins very distinct). Tibouchina fraterna N.E.Brown in McConnel & Quelch (1901: 28) differs from T. triprovincialis by the plane, non-bullate leaves with three veins, the two lateral running close to the margin (vs. three veins with the first pair of laterals equidistant between midvein and margins, plus a faint, incomplete pair), and thyrsoid inflorescences (vs. glomerulate). Both Tibouchina sipapoana Gleason (1950: 240) and T. steyermarkii Wurdack in Steyermark (1972: 873) differ from T. triprovincialis by the 1- nerved leaves (vs. 3-nerved plus a faint, incomplete pair of veins near the leaf margin). Tibouchina catharinae and T. striphnocalyx ( De Candolle 1828: 140) Gleason (1950: 241) differ from T. triprovincialis by the bigger leaves, paniculate inflorescences, and intersepalar sinuses with penicellate projections, these similar to the ones found in Pterolepis ( De Candolle 1828: 140) Miquel (1840: 72) (vs. glomerulate inflorescences, and intersepalar sinuses lacking penicellate projections). Finally, T. kunhardtii differs from T. triprovincialis by the sparse scales on the branches, and thyrsoid inflorescences (vs. branches densely covered with scales, and glomerulate inflorescences).

The specimen Anjos et al. 7 (RON) has been determined by P.J.F. Guimarães as Tibouchina aspera ; its label indicates duplicates in UPCB and RB, but these have not been located. Sanaiotti 274 ( US) was seen by the late John. J. Wurdack, who wrote that this specimen “was not matched at US ”, probably meaning that he already knew that it was an undescribed species.

Figures

FIGURE 1. Plate with images taken from the paratype and holotype of Tibouchina triprovincialis. A. Fertile branch. B. Stem with nodal and internodal scales. C. Leaf, abaxial surface. D. Leaf, adaxial surface. E. Apical glomerule. F. Flower bud. G. Pair of fused bracteoles, abaxial surface. H. Flower at anthesis, petals removed. I. Hypanthium and calyx, longitudinal section, abaxial surface. J. Antesepalous stamen. K. Antepetalous stamen. L. Gynoecium. (A, B, F from Silva et al. 23 at UPCB; C–E, G–L from Goldenberg et al. 3514, fixed in ethanol in the field).

FIGURE 2. Photos of Tibouchina triprovincialis in the field.A. Fertile plant. B. Stem and leaves profusely covered with scales. C. Flower at the beginning of the anthesis, with the stamens still folded inside the hypanthium. D. Branch apex with a glomerule.E, G. Flower, lateral view, in different positions. F. Stamens (AP: antepetalous; AS: antesepalous). H. Glomerule with young fruits. I. Branches with old fruits. (Photos by R. Goldenberg, from Goldenberg et al. 3514, Medeiros et al. 6337, and Mendes-Silva et al. 1190).

FIGURE 3. SEM of vegetative organs of Tibouchina triprovincialis. A. Internode. B. Node. C. Detail of the scales of the internode. D. Scales on the adaxial surface of the leaf blade and on the petiole. E. Adaxial leaf surface showing tector trichomes in grooves (zoomedin area indicated by dotted circle). F. Scales on the abaxial surface of the leaf blade and on the petiole. G. Detail of the abaxial surface showing scales and interspersed vermiform trichomes underneath. Scale bars: A, B = 0.5 mm; C = 50 µm; D = 0.5 mm; E = 250 µm; F = 1 mm; G = 50 µm. Zoomed-in areas in E = 100 µm. (From Goldenberg et al. 3514).

FIGURE 4. SEM of bracteole, hypanthium, andseeds of Tibouchina triprovincialis.A.Scales on the bracteole.B. Scales on the hypanthium and sepals (asterisks). C. Detail of the hypanthium showing scale ornamentation (zoomed-in area indicated by dotted circle). D. Seed showing surface detail (zoomed-in area indicated by dotted circle). Scale bars: A, B= 0.5 mm; C = 200 µm; D = 50 µm. Zoomed-in areas in C = 10 µm; in D = 25 µm. (A-C from Goldenberg et al. 3514, fixed in ethanol in the field; D from Capistrano et al. 114).

FIGURE 5. Occurrence map of Tibouchina triprovincialis.

FIGURE 6. A population of Tibouchina triprovincialis in the field. A. Grasslands at the “Parque Nacional dos Campos Amazônicos”, Machadinho D’Oeste, Rondônia, Brazil. B. Zoom from the same image, with red arrows showing how abundant the individuals of this species are among the grass. (Photos by R. Goldenberg).