Dichorisandra nana Aona & M.C.E. Amaral, 2014

Aona-Pinheiro, Lidyanne Yuriko Saleme, Bittrich, Volker & Amaral, Maria Do Carmo E., 2014, Two new species of Dichorisandra (Commelinaceae) from Rio de Janeiro and comments on the two species included in Vellozo’s “ Flora Fluminensis ”, Phytotaxa 184 (4), pp. 223-234 : 228-233

Dichorisandra nana Aona & M.C.E. Amaral , sp. nov. ( Figs. 3–5 )

This species appers to be close to D. thyrsiflora by its stamen morphology (6 yellow stamens, opening by a single apical pore), but diverges from it by its small stature, reaching only 30 cm (vs. 2 m tall in D. thyrsiflora ). Its androecium has the inferior abaxial stamen slightly erect, while the inferior lateral ones curved outwards and upwards, superior laterals curved upwards and the superior adaxial is also slightly erect, with a smooth to rugose ovary and whitish stripes on the adaxial surface of the lamina.

Type: — BRAZIL. Rio de Janeiro: Rio das Ostras, ca. 7.5 km S na rodovia para Armação de Búzios , 22°43’ S, 01°27’W, 25 m alt., 14 March 2004, fl., J . G . Jardim 4220, E . Lucas & S . Edwards (holotype: HUEFS!; isotype: CEPEC!, UEC!, RB!) .

Erect herbs, aerial stems unbranched, 10–30 cm tall. Stem glabrous to subglabrous at base, 2.5–3 mm diameter at base, 2 mm diameter at apex, internodes 6.8–7 cm long at base, shorter towards the apex. Leaves in rosettes or spirally arranged (internodes very short at the apex of the plant) or spirally-alternate, sheaths 2–2.3 cm long, glabrous, pilose only at the region opposite to the insertion of the lamina, margin ciliate, ciliae 1–2 mm long, whitish; petiole ca. 3mm to inconspicuous, glabrous; lamina ovate to elliptic, 12–16.5–21 x 5.5–6 cm, slightly discolorous, abaxial surface green, adaxial surface light green, young leaves with longitudinal white stripes on the adaxial surface, base asymmetric, apex acuminate, margin glabrous, lamina glabrous on both sides. Thyrse terminal, erect; basal bract leaflike; peduncle 1–1.5 cm long, sparsely short-pilose; peduncle of cincinni 1–2 mm long, 10–13 congested cincinni, each cincinnus 2–3-flowered, bracts of cincinni parallel to the cincinni, green becoming scarious, linear, 11–16 x 2–3 mm, shorter towards the apex, glabrous, margin glabrous, bracteoles scarious, 3–4 x 2–3 mm, glabrous, margin glabrous to sparsely ciliate at apex. Flowers bisexual or staminate, corolla actinomorphic, delicate, 1.5–1.8 cm diameter, pedicel 2–3 mm long, white, glabrous; floral buds ovoid to ellipsoid, 9–10 x 4–8 mm. Sepals elliptic, 10–12 x 4–8 mm, whitish with upper half purplish, glabrous; petals obovate to orbiculate, 10–13 x 10–11 mm, purplish to violet to purple with base whitish; stamens 6, subequal, pointing towards superior side of the flower, filaments 1–2 mm long, anthers sagittate, yellow, 5–6.5 mm long, opening by 1 apical pore, base cordate; ovary whitish, globose, glabrous, smooth to rugose, 1.5–2 x 1–2 mm, style 6–8 mm long, upper half pinkish to purplish, apex curved, stigma truncate; ovules 3–6/ locule. Capsule cylindrical, ca. 2 cm, epicarp rugose. Seeds ca. 1.5 x 1–2 mm, testa slightly rugose; aril inconspicuous in herbarium material ( Figure 4 ).

Distribution and habitat: —This species is known only from the “Região dos Lagos” in coastal Rio de Janeiro, more specifically for the municipalities of Armação de Búzios, Cabo Frio, Rio das Ostras and São Pedro da Aldeia. It can be found growing in marine influenced vegetation (‘restinga arbórea’) or in semi-deciduous Dry Forests (‘mata de encosta), between 0–300 m alt.

Phenology: —According to herbarium specimens, this species was collected flowering during January, February, March, May and October, and fruiting in March and December.

Etymology: —The epithet makes reference to the small stature of this species when compared to most of the remaining species of Dichorisandra .

Conservation status: —Endangered (EN B2 ab (ii, iii; IUCN 2001). Dichorisandra nana has a very restricted distribution (area of occupation: 20,78 km 2) and a small number of collections in two neighboring municipalities. Despite its scarcity, its habitats are in a rather remote area and we do not believe that it currently suffers from strong anthropogenic pressure besides some degradation of the habitat ( Figure 3 ).

Paratypes: — BRAZIL. Rio de Janeiro: Armação de Búzios, Mata de Encosta N voltada para a Praia de São João Fernandes , 25 May 2001, fl., C . Farney et al. 4372 ( RB); Cabo Frio , Morro do lado do Canal de Cabo Frio , fl., 17 January 1967, D. Sucre 1361 ( RB); Cabo Frio , Morro do Gavião , fl., 13 January 1968, D. Sucre 3936 ( RB); Cabo Frio , Morro do Gamboa , fl., 07 October 1968, D. Sucre 3796 ( RB); Cabo Frio , Morro do Mico , mata sobre solo argiloso, 19 Febuary 1997, fl., C . Farney & T. S . Perreira s.n. ( RB 424100); São Pedro da Aldeia , Serra de Sepiatiba , mata de encosta, 3 December 2001, fr., C . Farney et al. 4442 ( RB) .

Discussion: —This species resembles D. thyrsiflora due to the presence of six yellow stamens with anthers dehiscing by one apical pore. However, in D. nana the inferior abaxial stamen is nearly erect, the inferior lateral ones are curved outwards and upwards, the superior lateral ones are curved upwards and the superior adaxial one is also almost erect ( Figure 4 , 5 ), while in D. thyrsiflora there are four anthers positioned closely around the ovary with their apices connivent and two anthers in a lateral position, pointing outwards ( Figure 6 ). Furthermore, plants of D. nana are considerably smaller, reaching only ca. 30 cm, while D. thyrsiflora is a robust herb up to 2 m tall. Other differences between the species are the presence of white to silvery stripes on the upper blade of young leaves of D. nana , a character rarely observed in D. thyrsiflora , and the congested inflorescence of D. nana due to the short peduncles of the cincinni, 1–2 mm long (vs. 1–2 cm long in D. thyrsiflora ).

During the preparation of a taxonomic revision of the genus Dichorisandra ( Aona 2008) , we studied two names published by Vellozo in Flora Fluminensis (1829 [“1825”]. This Flora , which treats mainly plants from the state of Rio de Janeiro (and some from the neighboring state of São Paulo), is notorious for causing major headaches for taxonomists trying to establish the identity what Vellozo described. Herbarium specimens used by Vellozo when preparing the Flora were apparently completely lost (see Stafleu & Cowan 1986, Cervi & Rodrigues 2010) and the illustrations of the Flora Fluminensis are often chosen as lectotypes. In the absence of a text description, the illustrations can sometimes be considered as original descriptions of the new taxa, as many are “illustrations with analysis” (cf. ICBN Art. 38.8 and 38.9, McNeill et al. 2012). However, they are frequently of poor quality and are often very difficult to interpret .

Carauta (1969, 1973) discussed the problem of the effective dating of Vellozo’s work, relevant to establishing priorities for binomials published there. Since then, several of Vellozo’s names were synonymized ( Lima, 1995; Cervi & Rodrigues, 2010; Pastore, 2013). Cruz (1949) created an index of family names for Vellozo’s Flora, since Vellozo, following Linnaeus’ “sexual system”, had not used such names in his work. Sampaio & Peckolt (1943) undertook the task to list the names used by Vellozo (1829, 1831), accompanied by what they considered correct current names. Unfortunately, they did not discuss or justify their decisions. Their contribution must be considered of limited value, however, and may even have further added to the confusion existing over the names used by Vellozo. Taking into account the difficulties that even taxonomic specialists have in interpreting plants described in the Flora Fluminensis, this is hardly surprising.

Our analysis of Vellozo’s protologues (1825 [publ. 1829]: 142–143) and illustrations (Vol. 3, 1827 [publ. 1831]) resulted in the assignation of the two species of Vellozo’s Convallaria to the genus Dichorisandra (Commelinaceae) . The name of Convallaria diffusa Vell. , not yet typified, is herein lectotypified on the basis of the respective illustration in Vellozo’s Flora Fluminensis (1831). The second species, erroneously interpreted by Vellozo (1829) as C. racemosa L., is Dichorisandra thyrsiflora , herein also lectotypified. The basionym of Dichorisandra hexandra (Aubl.) C.B. Clarke , Commelina hexandra Aubl. is also lectotypified.

Convallaria diffusa Vellozo (1829 [´1825´]: 143), syn. nov. — Type: (presumed not to have been preserved or lost). Lectotype (designed here): Tab. 160, Fl. Flum. Vol. 3, 1827 [publ. 1831]. = Dichorisandra hexandra (Aubl.) C.B. Clarke. Bull. Torr. Bot. Club 29 (2): 703. (1992) Basionym: Commelina hexandra Aubl., Hist. pl. Guiane 1: 35, t. 12. 1775 (Type: Cayenne and Guiana, prope ripas fluviorum, Aublet s.n.,

s.d. Lectotype (designed here): BM!, isolectotype: P!).

Aublet (1775) described Commelina hexandra (1775: 35) based on the material he collected in Cayenne, French Guiana. The species was transferred to Dichorisandra by Schultes & Schultes f., who, in honor of Aublet, called it D. aubletiana (1830: 1181) . Kuntze (1891: 721) transferred it to the Stickmannia Necker (1790: 171) , using the correct specific epithet ‘ hexandra ’. This generic name, however, was not validly published by Necker, but only by Jussieu in 1827. Therefore it has no priority over Dichorisandra Mikan (1820: 1) . Clarke (1902: 703) made the new combination Dichorisandra hexandra giving the epithet ‘hexandra’ priority over ‘diffusa’.

Dichorisandra hexandra is usually of scandent habit, its flowers have purplish petals and six stamens. Other species of Dichorisandra also have these characteristics, and were frequently confused with D. hexandra . However, the material from Cayenne, deposited at BM and P, always has inflorescence axes covered with very short trichomes and dehiscent anthers with two apical pores. The species exhibits a wide variation in the blade size and pilosity, as well as in the stamen coloring. Nevertheless, such diagnostic characters as habit and anther dehiscence allow for the identification of herbarium material.

Sampaio and Peckolt(1943) synonymized Convallaria diffusa under D.pubescens Martius ex Schultes and Schultes f. (1830: 1186). Both species, Dichorisandra hexandra and D. pubescens , have six anthers dehiscent by two apical pores. However, D. pubescens can be differentiated from D. hexandra by its erect habit (vs. scandent habit with slender branches of D. hexandra ), by the uniform pilosity of its both blade faces (vs. the never uniform pilosity in D. hexandra ). The characters described by Vellozo (1825 [publ. 1829]) for Convallaria diffusa and especially the illustration (1827 [publ. 1831]) leave no doubt that it is D. hexandra . Therefore, since it is not possible to morphologically separate C. diffusa and D. hexandra , C. diffusa is herein synonymized under D. hexandra .

As regards the second name, Convallaria racemosa , we consider it used by Vellozo as misapplied (‘non auctorum’). The name was mistakenly applied to a taxon (‘ C. racemosa ’ sensu Vellozo ) distinct from that described by Linnaeus. Evidently, Vellozo did not intend to describe a new species, and therefore his C. racemosa cannot be considered a new synonym.

Convallaria racemosa Vellozo, Fl. Flumin. : 142. 1825 [publ. 1829], auct. non L.

= Dichorisandra thyrsiflora J.C.Mikan, Del. Fl. Faun. Bras. p. [7] e t. 3. 1820.

Type of D. thyrsiflora : (Type destroyed; other original material not found). Lectotype (designated here): Tab. 3, Del. Fl. Faun. Bras. 1820.

Convallaria racemosa was described by Linnaeus in´Species Plantarum´ (1753: 315–316) and subsequently transferred to the genus Maianthemum Wiggers (1780: 14) as M. racemosum ( L.) Link (1821: 343). Currently, Maianthemum is included in the Asparagaceae . Vellozo’s diagnosis (1825 [publ. 1829]) is identical to that of Linnaeus, but the former also includes in his “Observationes” various other characters: “stems erect, inflorescences terminal, corolla bluish, immature fruit with blue spots, roots tuberous”, among others. It is noteworthy that Maianthemum has white flowers (e.g., Frankie, 2014), as already mentioned in a short description of Plukenet (1696: 301) included by Linnaeus (1753) and apparently overlooked by Vellozo. Those of Dichorisandra thyrsiflora are bluish (as described and illustrated by Mikan 1820) and the fruits are stained dark blue in the upper part. These characters, together with the illustration of Vellozo shown in Tab. 160, Icones, vol. 3 (1831 [“1827”]), make it clear that we are indeed dealing with D. thyrsiflora , a species commonly found in the forests of Rio de Janeiro.

Sampaio & Peckolt (1943) synonymized Convallaria racemosa under D. gaudichaudinana Kunth (1843: 113) . However, D. gaudichaudiana has 4–6 anthers of rimose dehiscence, with the apex formed by the folding of external pollinic sacs, and with the abaxial blade face sparsely lanose, whereas D. thyrsiflora has 6 anthers of poricidal dehiscence and its abaxial blade face is glabrous.

Dichorisandra thyrsiflora can be characterized by its erect habit, thick stem, alternate spiraling leaves concentrated distally on the branch, and by the strongly zygomorphic flowers with 6 stamens and poricidally dehiscent anthers (with a single pore). Another striking feature is the arrangement of the lower lateral stamens that are widely divergent, pointing in the direction of the external side of the flower, a feature observed only in this species. Aona (2008), in her revision of the genus, notes that many floristic surveys mention D. thyrsiflora as a species present in several Brazilian states ( Barreto, 2005, Silva & Campos 2003, Rosary & Bianchini 2001). However, many of these identifications are incorrect, with D. thyrsiflora occurring mainly in the state of Rio de Janeiro in coastal vegetation with marine influence (‘restingas’), and only punctual collections in Bahia.

Mikan (1820), when describing D. thyrsiflora , does not mention the type material. The presumed type specimen of D. thyrsiflora in the Vienna herbarium of the Natural History Museum ( W) was most probably destroyed in World War II (http://www.nhm-wien.ac.at/en/research/botany/collections/lost_families). Seubert (1855) mentions collections of Mikan and Pohl of D. thyrsiflora from Rio de Janeiro, but without giving any details. It is well known, however, that the specimens of the Austrian scientific expedition to Brazil, in which Mikan and Pohl participated, were deposited in Vienna. As no duplicates of the respective collections of Mikan and Pohl are known, it has to be assumed that the original herbarium material used by Mikan for his description of D. thyrsiflora is lost. However, in the protologue Mikan provides a detailed description of D. thyrsiflora , and together with an illustration, the main features described above are represented. For these reasons, the illustration of the protologue of D. thyrsiflora is here proposed as the lectotype .

Figures

FIGURE 3. Map of distribution of. D. nana and D. odorata in Rio de Janeiro state. Note fragments of the Atlantic rain forest.

FIGURE 4. Dichorisandra nana. A. Habit. B. Hermaphrodite flower. C. Adaxial stamen. D. Lateral, superior stamen. E. Lateral, inferior stamen. F. Abaxial stamen. G. Gynoecium. H. Fruit. I. Seed, ventral surface with hilum exposed. J. Seed, dorsal surface, embryotega dorsal.All from Jardim 4220 (UEC). (originally from Rio das Ostras; cultivated at the UNICAMP).

FIGURE 5. Dichorisandra nana. Flower showing the position of the stamens.

FIGURE 6. Dichorisandra thyrsiflora. Flower showing the position of the stamens.