Vellozia dracaenoides R.J.V. Alves & N.G. Silva., 2014

Vellozia dracaenoides R.J.V. Alves & N.G. Silva. View in CoL , sp. nov.

Type:— BRAZIL, Minas Gerais: Município de Ouro Branco, Serra do Ouro Branco, northern slopes of summits , on quartzite and itabirite outcrops, 1356 m, 20°29’19.68”S, 43°41’20.04”W, 10 January 2006, R.J.V. Alves et al. 7615 (holotype R!) GoogleMaps .

Diagnosis:— Vellozia dracaenoides is close to V. compacta , from which it differs by taller stature, sparingly branched crown, hypanthium tube longer than the ovary, lack of staminal appendages, leaf vascular bundles with aquiferous hypodermis abaxially and adaxially, and the abaxial sclerenchyma sheath twice the size of the adaxial.

Solitary dracenoid plants to 360 cm tall with evident vegetative reproduction by rooting from old nodding branches; Caudex sparingly (2–9) branched; Phyllotaxis spirotristichous; Leaf sheaths lustrous brown with pale yellow serrulate margins, abscission zone truncate, older withered sheath apices pectinate and cinereous by exposure of vascular traces; Leaf laminas soon deciduous, straight, linear-triangular, conduplicate, 14–35 × 0.8–1.7 cm, glabrous, margins densely serrulate throughout; Bract narrowly triangular, 5.5 cm long; Flowers 1–3 per rosette; Pedicels much shorter than the leaves, longitudinally grooved, trigonous, 5–15 cm long, 3 mm in diam., smooth or with emergences on the apical half; Hypanthium obconical, 1.7–3.0 cm long, 1.1–1.2 cm in diam., at anthesis surface sticky-resinous, sometimes verrucose (with small, rounded to conical glandular emergences); Hypanthial tube exceeds the length of the ovary by 1–2 mm; Tepals at anthesis perpendicular to flower axis, widely ovate-elliptic, 6.0–7.0 cm long, 2.0– 3.2 cm wide, basal ¼ pale violet, remainder purple; Stamens ~36, with violet, staminal appendages not found, Filaments 10–12 mm long, yellow; Anthers 10–14 mm long, yellow; Style 3–4 cm long, yellow; Stigma 6–7 mm diam., trilobate, yellow. Capsule terete-trigonous, 1.4–1.7 cm in diam., 2.2–2.5 cm long, glabrous; ( Fig. 1–3 View FIGURE 1 View FIGURE 2 View FIGURE 3 ).

Distribution and habitat: —Known only from three populations, all on the northern flanks of the summits of the Serra do Ouro Branco range in Minas Gerais State, among quartzite and itabirite outcrops of campo rupestre. V. compacta and V. dracaenoides are sympatric in the sense that they occur in the same mountain range, but the populations of V. compacta are concentrated on the purely quartzitic outcrops of the southern ridge and summits, while those of V. dracaenoides occur on a transition of quartzite and itabirite outcrops on the northern slopes and along northward, sub-summit drainages. The populations of both species were not found in direct contact.

Etymology: —The epithet invokes the vegetative convergence of this large Vellozia species with shrubby members of Dracaena Linnaeus (1767: 229) (Asparagaceae) .

Paratype: — BRAZIL, Mun. Ouro Branco, Serra do Ouro Branco , 1425 m, 20 o 28’48”S, 43 o 43’08”W, 11 December 2007, R.J.V. Alves 8283 ( R!) GoogleMaps .

Notes: — Vellozia dracaenoides shares the densely imbricate leaf arrangement and brown but apically cinereous leaf sheaths with V. gigantea ; evenly truncate leaf sheath apices with V. glabra ; a straight (non-arcuate) leaf lamina with all further species of the dracenoid group; oblong to obtriangular hypanthia and capsules with V. spiralis and V. compacta and, with the latter, also the soon deciduous straight leaf blades.

Compared leaf anatomy: —Surface view of epidermis: Both surfaces in both species are glabrous. In V. dracaenoides abaxial cells mostly rectangular, anticlinal thin-walled; adaxial cells rectangular to square-shaped but with slightly thicker anticlinal walls (3A–B); In V. dracaenoides the stomata are sunken in both surfaces, mostly anomocitic, some paracitic or tetracitic ( Fig. 3A–B View FIGURE 3 ), more numerous on the adaxial surface (in V. compacta , more numerous on the abaxial surface).

Transverse section of lamina: Both species have dorsiventral, conduplicate laminas; Adaxial surface smooth; abaxial surface furrowed almost to half blade thickness in both species ( Fig. 3C View FIGURE 3 ). Epidermis cells more or less rounded in both surfaces vs. abaxial square shaped in V. compacta . Cuticle thickened in both faces in both species ( Fig. 3C View FIGURE 3 ). Leaves are amphistomatic with stomates concentrated mostly in the furrows in both species ( Fig. 3F View FIGURE 3 ). The mesophyll is compartmentalized by the vascular bundles and by the furrows connected with translucent parenchyma and the latter with the adaxial sclerenchyma bundles. In V. dracaenoides the mesophyll has 1–2 layers of translucent parenchyma cells, interrupted in the adaxial surface by sclerenchyma bundles strictly above the furrows, while V. compacta has three layers of parenchyma interrupted by numerous short sclerenchyma strands ( Fig. 3C View FIGURE 3 ). Adjacent to these layers, V. compacta has 1–2 additional layers of translucent parenchyma cells. V. dracaenoides has 5–6 layers of palisade parenchyma ( Fig. 3C–D View FIGURE 3 ) vs. 3–4 layers in V. compacta . The remainder of the mesophyll consists of many layers of compactly arranged spongy parenchyma in both species. In both species there are 5–6 rows of large translucent cells above the furrows ( Fig. 3C–D View FIGURE 3 ), and in V. compacta 3–4 rows of these cells also occur adaxially to the bundles. The vascular system in both species has 2 phloem strands abaxially flanking the xylem and separated by parenchyma cells ( Fig. 3E View FIGURE 3 ). In V. dracaenoides the abaxial sclerenchyma cap is larger than the adaxial and is Y-shaped ( Fig. 3C View FIGURE 3 ), while in V. compacta it is smaller and V-shaped. A distinct bundle sheath completely envelops the vascular bundles in both species; however, in V. dracaenoides it extends towards both faces of the epidermis by 1–2 additional parenchyma cell layers ( Fig. 3C View FIGURE 3 ). The leaf anatomic characters of both collections of V. compacta are identical.

Chemistry:—Screening of the dichloromethane extract from Vellozia dracaenoides revealed 14 diterpenes. Two diterpenes were previously described from V. compacta : CLEISTHANTA-8,11,13-TRIEN-7-ONE ( Riehl & Pinto 2000), also isolated from Vellozia aff. caruncularis Martius ex Seubert in Martius (1847: 78) ( Silva et al. 2001) and from V. flavicans Martius (1829: 293) ( Pinto et al. 1997); ISOPIMARA-8(9),15-DIENE-7,11-DIONE from V.compacta ( Riehl & Pinto 2000) and from V. epidendroides Martius (1829: 292) ( Pinto et al. 1990); CLEISTHANTA-6,8,11,13- TETRAENE known from V. declinans Goethart & Henrard (1937: 369) ( Pinto et al. 1992); and ISOPIMARA-8(9),15- DIENE-7- ONE. This last substance had not previously been isolated from a Vellozia , having been synthesized by Pinto et al. (1979) and later isolated from Salvia parryi Gray (1870: 369) by Touche et al. (1997). The other 10 substances could not be identified based on mass spectrometry comparison with our data base and scientific literature.

Discussion: —Which populations represent the real Vellozia compacta? The original diagnosis is relatively vague: (thickened caudex; ensiform, erect, rigid leaves with truncate-obtuse apex and acutely serrulate keel and margins; obovate 3-agulate capsule with margins basally verruculose). Moreover the type locality was not precisely indicated ( Brazil, 1 January 1817, sine loc., C. Martius 725, holotype P, isotype MO). If only the collection date on the type were correct, this could pinpoint the topotypic population, but Martius only set foot in Brazil on July 15 th of that year ( Spix & Martius 1824). The Field Museum of Natural History (F) holds a type with unknown status: Brazil, 10°51’57.96”S, 52°52’23.16”W, C.F.P. Martius s.n. but the coordinates were inferred in an unclear manner and point to a forest near an urban area in the State of Mato Grosso, which is neither a typical Vellozia habitat nor was visited by Martius.

Based on 149 collections using the binomial Vellozia compacta , a map generated by SpeciesLink (2014) indicates a range of occurrences spanning ca. 430 km, mainly on the north-south axis in the State of Minas Gerais. This broad distribution is uncommon for the genus, in which most species have small populations and tend to be narrowly endemic ( Giulietti & Pirani 1988, Giulietti et al. 1997), often being known only from their holotypes ( Smith & Ayensu 1976). One possible explanation for the assumed broad distribution is the misuse of the binomial, including several species that may be morphologically difficult to distinguish. The morphological polymorphism and high genetic divergence among populations reported by Lousada et al. (2013) seem to support this assumption.

The use of morphology, leaf anatomy, chemical constituents, phytogeography together with the findings of Lousada et al. (2013) from molecular data may be the only way to resolve this complex and shall probably reveal several previously unrecognized “cryptic” species. The populations studied by the latter authors have a strong geographic congruence, with most spatially close populations showing more genetic affinity.

All evidence suggests that there are many cryptic species under Vellozia compacta . The “clear variations in sclerenchyma patterns” found by Ayensu (1974) and other anatomical divergences found in the course of this study in specimens determined as V. compacta in major herbaria also suggest the need for further studies. Ayensu (1974) studied Irwin 2407 (NY, US) from Serra da Piedade (Minas Gerais) and Maguire et al. 49132 (NY, US), Segadas-Vianna & Loredo 1085 ( US) and L. B. Smith 7005 ( US) from 3-5 km E of Serra on road to Diamantina (Minas Gerais). In subsequent molecular work by Lousada et al. (2013), the Diamantina and Serra da Piedade populations appear in two major separate groups that most probably represent distinct taxa.

Insofar, chemical constituents attributed to V. compacta were sampled from several localities. For instance Pinto (1985) obtained material in the Cipó Range while Riehl & Pinto (2000) obtained theirs from Diamantina. Considering the geographic distances among these populations, and the fact that populations from those areas differ on the molecular level ( Lousada et al. 2013), the chemical constituents described from each population may pertain to distinct species. Mello-Silva (2000, 2004) placed V. compacta in a group which shares spirotristichous phyllotaxis, staminal appendages, hypanthium tube shorter than the ovary, and aquiferous parenchyma extending adaxially above both the abaxial furrows and the vascular bundles. In V. dracaenoides staminal appendages were not found and the hypanthial tube exceeds the ovary. In Smith & Ayensu (1976), V. compacta keys out in two distinct parts, “p. 41: Flowers erect, not exceeded by the leaves” and “p. 42: leaves exceeding the flowers”. Hence the flowers at anthesis in V. compacta may or may not exceed the leaves in each rosette whereas in V. dracenoides they open deep within the rosettes ( Fig. 2B View FIGURE 2 ). The term “dracenoid group of Vellozia ” was coined by Mello-Silva & Menezes (1999), circumscribing the six species mentioned in the introduction, all of which stand out by their relatively large stature and by the following characters: Stems thickened at the base with leaves apically concentrated and soon deciduous, leaving persistent leaf sheaths; flowers long-pedicellate with violet tepals; numerous stamens (at least 18); conspicuous staminal appendages; capsules with apical dehiscence; fibro-vascular bundles with two phloem strands; and aquiferous hypodermis extending abaxially to bundle-sheaths and furrows (the last character does not occur in V. dracaenoides ).

Tall stature alone may have been selected for in several parallel lineages of Vellozia , and thus may not be a useful taxonomic character. Several sterile populations of Vellozia taller than 2(4–5) m have been found in many localities of Bahia, Goiás, Mato Grosso, Minas Gerais and Tocantins. It is very probable that many such populations represent new species that still need to be properly sampled and described.

Subsequently, in morphological analyses of limited numbers of species, Mello-Silva (2004, 2005) associated V. compacta with other taxa, none of which include the aforementioned dracenoid species. Of all these, a recent treatment ( Mello-Silva et al. 2011) did not include V. obtecta Mello-Silva (2004: 453) , V. peripherica Mello-Silva (2004: 457) , and only included V. compacta from the original “dracenoid group”. The authors claim to have solved the generic delimitations within a very broadly defined Velloziaceae using only 20% of the taxa in a “total-evidence” analysis. It is not clear how total evidence can be obtained with only 20% sampling.

Most species of Vellozia have very narrow distribution ranges. The identities of the host plants from two localities over 1,100 km apart, one in the State of Tocantins and another in Minas Gerais, both claimed in a study of endophytic fungi by Rodrigues (2010) to be V. compacta , almost certainly represent a misidentification of the host species. Furthermore, the author did not deposit voucher specimens of the host plants for his studies, instead electing a specimen (Badini 4145, OUPR) from Itacolomi peak near Ouro Preto, which is a totally different locality. The same applies to a collection plotted on a map in Mato Grosso state by SpeciesLink (2014).

Hence it seems that the “Dracenoid group” of Vellozia , as proposed by Mello-Silva & Menezes (1999), is either artificial or in need of a more detailed circumscription. Thus, size alone may not be a consistent character for the segregation of a “dracenoid” group, nor the separation of Vellozia spp.

Vellozia dracaenoides exhibits conspicuous vegetative propagation in which the subapical parts of drooping branches, when damaged by wind or other factors, touch the ground and new roots establish a clone which eventually becomes an independent genet. For instance, Mello-Silva & Menezes (1999) reported this type of reproduction for V. auriculata . We have also observed this in rare individuals of V. compacta in the Ouro Branco and Bico de Pedra ranges.

Conservation issues: —Lumping many isolated populations under a single binomial may be deleterious to conservation efforts ( Morrison III et al. 2009). In recent decades, a few unresolved species complexes have received the designation of “ochlospecies” (sensu Cronk 1998). This concept was recently applied to Vellozia hirsuta Goethart & Henrard (1937: 374) by Barbosa et al. (2012). However, it seems reasonable to assume that in Velloziaceae such a concept is only the result of insufficient sampling. Thus, additional detailed studies of all available populations combining morphology, anatomy and molecular data would split such complexes into separate species.

Many populations of the tall Vellozia species are currently at risk of extinction due to various causes, e.g. habitat loss due to mining of iron ore (e.g. Carmo & Jacobi 2013. Jacobi et al. 2007, Jacobi & Carmo 2012, Vincent & Meguro 2008), small natural populations and exploitation by local peoples ( Alves 1994). Even though the populations of V. dracaenoides in the Ouro Branco range are officially (but not effectively) protected in a State Park established in 2009, the population size and Area of Occupancy (AOO) are small. Less than two hundred individuals of this species grow in a mosaic with a total AOO of less than 1 km 2. No juvenile individuals are known and the old, mature plants are being decimated by frequent fires. Hence, V. dracaenoides is to be classified as Critically Endangered [CR 2ab(ii, iii)+C2a(i)] following IUCN (2012) Red List Categories and Criteria.