Synopsis of Commelina L. (Commelinaceae) in the state of Rio de Janeiro, reveals a new white-flowered species endemic to Brazil

Abstract A synopsis for the genus Commelina in the state of Rio de Janeiro, Brazil, is presented here, including a new species, ten new synonyms, five designated lectotypes, two designated epitypes and an excluded name. Commelina huntii, a new species, is remarkable due to the combination of rusty to rusty-brown hairs at the margin of its leaf-sheaths, connate spathes, white flowers with auriculate medial petal, ovaries with sparse black papillae and dehiscent fruits. Additionally, we provide an identification key, illustrations, and conservation status for the species of Commelina recorded in the state of Rio de Janeiro.


Introduction
Commelina L. is the largest genus of Commelinaceae, comprising between ca. 170 species (Faden 1998) and 205-215 species (Govaerts and Faden 2009;The Plant List 2013, respectively). It is one of the six genera of Commelinaceae (out of 42) to have a cosmopolitan distribution (Faden 1998), and one of the most complicated taxonomically. Commelina is easily differentiated from the remaining genera in the tribe by its inflorescences which are subtended by spathaceous basal bracts and reduced to (1-)2 fasciculate cincinni, zygomorphic flowers, petals clawed, unequal and mostly blue (but sometimes white or lilac, rarely yellow, apricot or orange), three posterior staminodes with 6-lobed cruciform antherodes, three anterior stamens, and 2-locular or unequally 3-locular and 2-valved capsules (Faden 1998).
The state of Rio de Janeiro is entirely placed within the Atlantic Forest domain (IBGE 2012), being one of the four diversity centers of the family, and possessing 67 of the 96 accepted Commelinaceae species for Brazil (BFG 2015). With 1,109,546 ha of continuous forests, which represent 7% of what remains of the Atlantic Forest, Rio de Janeiro is also the state with the greatest amount of preserved forest remnants from this biome (Ribeiro et al. 2009). Nevertheless, the most recent published state flora is nearly two centuries old (i.e. Vellozo 1829) and no satisfactory taxonomic treatment for the Commelinaceae has been done since. As a first attempt to clarify the taxonomy of Commelina in the state of Rio de Janeiro, we describe a new species of Commelina, together with a synopsis for the genus in the state. This work includes an identification key, illustrations and an overview of some overlooked Brazilian Commelina names.

Methods
The descriptions and phenology of the species were based on herbaria, spirit, fresh material, field data, and literature. All species were studied in the field and had their descriptions complemented with field notes, photographs, and cultivated specimens, gathered between the years of 2010-2016. Live specimens collected by the authors were kept in cultivation at the greenhouse of the Jardim Botânico do Rio de Janeiro, in order to better observe, photograph and analyze fresh flowers, fruits and seeds, as well as other phenological data. Specimens from the following herbaria were also analyzed: ALCB, BHCB, BHZB, BM, BOTU, CEPEC, CESJ, CGE, CNMT, CVRD, ESA, FCAB, FLOR, FURB, GUA, HAMAB, HAS, HB, HBR, HRB, HSTM, HUEFS, HURB, IAC, ICN, JOI, K, MBM, MBML, PMSP, R, RB, RFA, RFFP, SP, SPF, UEC, UPCB, and US (herbaria acronyms according to Thiers, continuously updated). While specimens of the following herbaria were analyzed using high-resolution images available on-line: B, BRIT, C, CAL, F, INPA, L, MG, MO, MY, NY, P, U, and WAG. The classification of the vegetation patterns follows IBGE (2012). The indumenta and shapes terminology follows Radford et al. (1974); the inflorescence terminology and morphology follows Weberling (1965Weberling ( , 1989 and Panigo et al. (2011); the fruit terminology follows Spjut (1994) and Joseph & Nampy (2012); and seeds terminology follows Faden (1991) and Joseph & Nampy (2012). The conservation statuses were proposed following the recommendations of IUCN Red List Categories and Criteria, Version 3.1 (IUCN 2001). GeoCAT (Bachman et al., 2011) was used for calculating the Extent of Occurrence (EOO) and the Area of Occurrence (AOO). The typification of Vellozo's names for Commelina followed the same methodology used by Pellegrini (2015), Pellegrini et al. (2015) and Pellegrini & Carvalho (2016).

Results
Key to the species of Commelina in Rio de Janeiro state, Brazil Distribution and habitat. Tropical and subtropical regions of the world. In the state of Rio de Janeiro it is especially common in disturbed areas of drier regions inland or near the coast, being common in restingas (i.e. sandbank vegetation), and as a weed in agricultural fields (Fig. 2).
Phenology. Throughout the year, but especially in the rainy season. Conservation status. It is a weed of worldwide distribution being very common in sunny disturbed areas and in agricultural fields. Following the IUCN recommendations (IUCN 2001), it should be considered as Least Concern (LC) in the state of Rio de Janeiro and worldwide.
Morphological and ecological notes. The underground cleistogamous flowers and fruits seem to be produced only in areas where the soil is soft. The flower morphology differs from the aerial chasmogamous in pigmentation (being paler), while the fruits are sub-globose (due to larger seeds).    Distribution and habitat. Tropical and subtropical regions of the world, being very common in shady disturbed areas such as road sides, gardens and forest margins, and in agricultural fields. It is also found growing on the edge of wet paddy fields, ponds, ditches and stream sides (Fig. 2).

Commelina diffusa
Phenology. Throughout the year, but especially in the rainy season. Conservation status. Following the IUCN recommendations (IUCN 2001), it should be considered as Least Concern (LC) in the state of Rio de Janeiro and worldwide.
Morphological notes. The specimens from the state of Rio de Janeiro usually show a staminode malformation, i.e. the central antherode is lacking in most of the herbaria and living specimens examined.
Nomenclatural and taxonomical notes. Commelina communis Vell. is a later homonym of C. communis L. (the genus' type species), thus, rendering it illegitimate. Vellozo's description (1929) is little informative, especially for Commelina, lacking all the characters evidenced by Faden (2008) as important to delimitate species in the genus. Despite this, the spathe and flower details (Vellozo 1831: t. 75), along with the leaf shape and stem diagnosis (Vellozo 1929), make it possible to associate C. communis Vell. to C. diffusa, rather than to C. deficiens Herb. (= C. erecta), as pointed out in the Index Methodicus of Flora fluminensis (Vellozo 1831, v. 1). Distribution and habitat. Tropical and subtropical regions of the world, being common in disturbed areas of drier regions inland or near the coast, commonly found in restingas or in urban areas (Fig. 2).

Commelina erecta
Phenology. Throughout the year, but especially in the rainy season. Conservation status. Following the IUCN recommendations (IUCN 2001), it should be considered as Least Concern (LC) in the state of the Rio de Janeiro and worldwide.
Nomenclatural and taxonomical notes. Clarke (1881), in his revision of Commelinaceae, erroneously considered C. erecta as a synonym of C. virginica L., a species endemic to the USA (Faden 2000). Thus, some names currently placed under the synonymy of C. erecta were originally described under C. virginica, or transferred to it at some point. According to Faden (1993Faden ( , 2000, Commelina erecta can be differentiated by its leaf-sheaths with auriculate margins, medial petal linear and hyaline, and all locules 1-seeded (vs. leaf-sheaths not auriculate, medial petal blue and conspicuous, and dorsal locule 1-seeded and ventral locules 2-seeded, in C. virginica).
There seems to be some doubt regarding C. deficiens Herb. synonymy. According to Tropicos.org (2015), this species is considered a synonym of C. erecta. Nevertheless, eMonocot.org (2010) and The Plant List (2013) treat C. deficiens under the synonymy of C. virginica. As abovementioned, there is an historical confusion regarding C. erecta and C. virginica. If we exclusively take into account that C. deficiens was described by Herbert (1826) from the surrounding areas of Rio de Janeiro, it is impossible for C. deficiens to be conspecific to C. virginica. Added to that, the watercolour presented by the author perfectly illustrates the habit, flower morphology and the inflorescence characteristic of C. erecta (with the aborted upper cincinnus). Thus, there is little doubt that C.
deficiens is a synonym of the latter. According to Stafleu and Cowan (1979), Herbert's type specimens were deposited at K, but no specimen corresponding to C. deficiens was found. Thus, in accordance to the Code (McNeill et al. 2012, Art. 9.12), in the absence of herbarium specimens, we designate this illustration as the lectotype for C. deficiens.
Commelina villosa (C.B.Clarke) Chodat & Hassl. has long been a name of dubious application. Clarke (1881) had already noticed that its morphological concept overlapped with the one of the highly variable C. erecta, and that the difference between them relied solely on the plant's indumenta. The observation of a great number of natural populations and specimens kept in greenhouses showed that most of the morphological variation known for C. erecta has an environmental background. Large flowered specimens developed into small flowered specimens after being transplanted from sunny to shady areas. The same thing happened to narrow-leafed and erect plants (which would represent C. erecta var. angustifolia), developing into creeping and small to wide-leafed plants. The indumenta also varied when specimens were transplanted from the field to the greenhouse. Regarding growth form and position of the stem of C. erecta, the plants can vary from creeping to sub-scandent (i.e. stems leaning generally on bushes or any other kind of support) to partially or completely erect. The only morphological characters, constant in all areas and environmental conditions were: the auriculate leaf-sheath margins; terminal to apparently terminal inflorescences (1-3 per stem), broadly sagittate to subcordate spathes with connate margin, aborted upper cincinnus (generally completely absent, but sometimes only vestigial); hyaline, linear and involute medial petal (almost invisible at blind sight); capsules with 1-seeded locules; and smooth seed testa.

Commelina huntii
Specimens seen ( Etymology. This species is named after the British botanist Dr. David R. Hunt, in honor of his extensive contribution to Commelinaceae systematics worldwide, especially for his contributions to Tradescantieae and the "Phaeosphaerion group" of Commelina. Distribution and habitat. Commelina huntii was collected in moist and shaded nebular forests, generally near water bodies, in the states of Minas Gerais, Rio de Janeiro, and São Paulo, in elevations from 800 to 1,700 m above sea level (Fig. 5). In very rare cases it can also be found in open sometimes disturbed areas.
Phenology. It was found in bloom from November to June and in fruit from December to March, rarely in June.
Conservation status. Despite the wide EOO (112,904.528 km 2 ), the AOO (40.000 km 2 ) is considerably small, since all known populations are significantly small and fragmented. Following the IUCN recommendations (IUCN 2001), C. huntii should be considered as Endangered [EN,B2b(ii,iii)c(iv)+C2a(i)] in its overall distribution.
Affinities. Commelina huntii can be recognized by its white flowers with auriculate medial petal and sparsely papillose ovary and capsule. It is similar to C. rufipes Seub. due to its white flowers and rusty hairs on the leaf-sheaths, but it can be readily distinguished from the latter by its connate spathe base (vs. free base); auriculate medial petal without medial constriction (vs. entire medial petal with medial constriction); light-brown, ellipsoid, dehiscent capsules (vs. pearly-white to silvery, globose, crustaceous capsules); and by its free, ornamented seeds (vs. seeds fused to the capsule septa, forming a dispersal unit, with smooth testa). The gross floral morphology of C. rufipes is much more similar to the C. benghalensis than the one of C. huntii, possessing only the white petals in common.
Commelina huntii is most similar to C. obliqua Vahl due to its oblique leaf blades, connate spathe base, dehiscent capsules, and ventral seeds free with foveolate testa. Despite this, C. huntii can be distinguished from C. obliqua by its densely setose leafsheath margins with rusty to rusty brown hairs (vs. leaf-sheath margins long-ciliate with light to medium to dark brown to atro-vinaceous hairs); petals white (vs. blue to light blue to lilac to pale lilac), paired petals broadly rhomboid to rhomboid reniform (vs. broadly ovate to broadly ovate reniform), medial petal cucullate and biauriculate (vs. involute and entire); anthers of the lateral stamens light-yellow to cream with margins tinted vinaceous (vs. completely orange); ovary and capsules sparsely black papillate (vs. smooth); 1-2 capsules per spathe (vs. 5-7); seeds with peach-colored farina (vs. seeds white farinose), and dorsal locule seeds with shallowly foveolate testa (vs. rugose foveolate testa). Distribution and habitat. Mexico to Argentina being very common in shady disturbed areas such as road sides, gardens and forest margins, and in agricultural fields. It is less commonly found growing in drier regions and rocky outcrops (Fig. 2).

Commelina obliqua
Phenology. Throughout the year, but especially in the rainy season. Conservation status. Following the IUCN recommendations (IUCN 2001), as currently circumscribed, C. obliqua should be considered as Least Concern (LC) in the state of Rio de Janeiro and worldwide.
Morphological notes. A great deal of morphological variation can be observed in C. obliqua and in its current circumscription. It comprises plants from small to large stature (sometimes more than a 1.5 m high); stems from creeping with ascending apex to erect to sub-scandent, and thin and fibrous to robust and somewhat succulent stems. The leaves can vary from 4-20 cm long, from glabrous to scabrous to pilose, and from green to dark green to vinaceous abaxially. Flower size and color also vary, which as in C. erecta seems to be environmentally influenced, probably by differences in soil pH and light intensity (Pellegrini, pers. obs.). The petals of C. obliqua can range from intense blue to light blue, sometimes varying from lilac to pale lilac. Commelina obliqua likely represents a species complex and biosystematic studies are necessary in order to better understand and elucidate its boundaries. Until this is addressed we believe that a wide circumscription, as presented here, is currently the best way to deal with this taxon. forests in the Cerrado biome. It is a rare species in the Atlantic Forest and Cerrado biomes, with most of its collections being from the Amazon Forest.

Commelina rufipes
Conservation status. As abovementioned, C. rufipes is locally rare in the state of Rio de Janeiro and not as frequent in the field as the blue flowered species of the genus. Despite its wide distribution, it seems to occur only in preserved rainforests, forming dense but isolated populations. Data regarding its reproductive cycle would be of great value for this species' conservation. Following the IUCN recommendations (IUCN 2001), C. rufipes as currently circumscribed should be considered as Least Concern (LC) in its worldwide distribution.
Taxonomical notes. We currently accept two varieties within this species (sensu Faden & Hunt 1987). The floral morphology of both varieties of C. rufipes is poorly understood as little reproductive material exists. However there seems to be no morphological overlap in vegetative characters and very little overlap in their distributions. Further biosystematic study, focusing especially on floral morphology, would be most useful in evaluating their boundaries and taxonomic status. Commelinopsis glabrata D.R. Hunt, (1981: 195 Nomenclatural notes. When describing C. rufipes, Seubert (1855) only mentions that his new species was based on a Martius specimen, at M. After searching the M collection, we found just two specimens from this collector -Martius 76 (M0210921) and Martius 77 (M0210920). Since the specimen Martius 76 was clearly annotated in Seubert's handwriting it is the obvious choice for a lectotype. Nonetheless, Seubert's original description makes it clear that all available specimens had few if any flowers, which matches the specimens found by us at M. This has caused great taxonomic problems over the years, with this name being ascribed to a number of different genera (i.e. Athyrocarpus Schltdl., Commelina, Commelinopsis Pichon, and Phaeosphaerion Hassk.), and as either accepted or as a synonym by different authors (Faden & Hunt 1987). Thus, in accordance to the Code (McNeill et al., 2012, Art. 9.8), we also designate a well-preserved flowering specimen as an epitype, to avoid further taxonomic and nomenclatural problems.
Taxonomical notes. Apart from the obvious difference in indumenta, the leaves of C. rufipes var. rufipes tend to be thinner (lanceolate to elliptic-lanceolate), with a cuneate base and acute apex, while the leaves of C. rufipes var. glabrata tend to be wider (ovate-elliptic to ovate), with a round to obtuse base and acuminate apex.

Excluded name
After a thorough analysis of Vellozo's description and original illustration for C. singularis Vell., it became clear that this species does not belong in the genus Commelina. This name is better placed under the synonymy of Tripogandra diuretica (Mart.) Handlos (Syn. nov.), and the necessary taxonomic and nomenclatural comments and typifications are made below.  Taxonomical notes. Vellozo's plate for C. singularis (1831: t. 76) shows a creeping plant with eudicot-like leaves (net-veined and a single apparently trifoliate leaf), not identifiable as any known species of Commelinaceae. Nevertheless, the inflorescence type (Fig. 6A, C), details of the androecium (Fig. 6B, D), and the morphological description of six dimorphic stamens, three of which are staminodial -"Stamina sex, quorum tria nectaria mentiuntur" -(Vellozo 1829), consistently allows this name to be associated to the genus Tripogandra Raf. Another remarkable feature of Vellozo's plate is the gynoecium, which is illustrated with a very short and slightly curved style (Fig. 6B). This feature distances C. singulars from the genus Commelina where the style is long and sigmoid, bringing it closer to Tripogandra. The leaves illustrated by Vellozo belong to the genus Polygonum L. (Polygonaceae), which usually possesses white to pink to lilac flowers, and occurs in the same marshes as T. diuretica. This confusion is apparently common in Brazilian herbaria, where Polygonum specimens are commonly misidentified as commelinaceous taxa (Pellegrini pers. obs.).

Commelina singularis
Vellozo (1829: 31) also mentions that C. singularis is found growing in slow-water environments -"Aquis stagnantibus, et confluentibus habitat" -. Only T. diuretica and T. warmingiana (Seub.) Handlos occur in the state of Rio de Janeiro; the first being very common, extremely variable in size and flower morphology, and normally occurring in marshy areas; the second being very rare, uniformly small in size and flower morphology, and occurring in drier areas (Pellegrini et al. 2013). Thus, C. singularis is here regarded as a synonym of T. diuretica. In accordance to the Code (McNeill et al., 2012, Art. 9.8), in order to avoid future confusions and to fix the application of this name, we herein designate an epitype for C. singularis.

Discussion
Our work has reaffirmed the importance of thorough descriptions, fieldwork, photographs, spirit collections, and cultivation of specimens to better understand the taxonomy and systematics of intricate genera such as Commelina. This genus in particular poses problems as floral characters are difficult to observe in dried specimens (e.g. Faden 1993Faden , 2008Faden , 2012Nampy et al. 2013), which calls for particular attention to be paid to adequate description of these in any new species (Faden 2008). Historically there are examples where either floral, fruit, or seed characters are only incompletely described, or even omitted. In some cases, the available specimens possess such strikingly different habit or vegetative characters, that the name can be easily applied (e.g. Faden et al. 2009). Nevertheless, in most cases, the lack of appropriately detailed description may cause confusion or prevent identification and the application of a correct name. Capturing the range of a species' phenotype is also important and population studies have shown to be of great use, especially in the Neotropical species, allowing us to record and compare wide ranges of morphology. The description of new taxa, based on few and odd specimens needs to be carefully considered, and is a strategy that tends to inflate the description of unnecessary or invalid new species and infraspecific taxa.
Characters such as inflorescence position and morphology, spathe shape and conation, petal and fruit morphology, and seed ornamentation play important roles in species distinction and delimitation, in Commelina. Nevertheless, characters once thought to be useful in species delimitation such as indumenta and leaf shape have shown to be highly variable within the same species and thus not completely reliable. This is easily observed in all Neotropical species, and most of the wide-spread species (e.g. C. benghalensis, C. diffusa, and C. erecta). Growth-form and subterraneous system morphology are also potentially interesting for the taxonomy of Commelina worldwide. On the other hand, most of the morphological characters pointed out by previous authors (e.g. Faden 1993Faden , 2008Nampy et al. 2013) as key to the taxonomy of the genus are mostly difficult to observe in herbaria specimens. Thus, work to expand and refine the morphological tools available to workers in this group should be ongoing. It is also apparent that some species still need further systematic study in order to clarify their boundaries. The Pantropical C. diffusa complex is poorly understood in the Neotropical region and is probable that more than one species, being treated under the widely polymorphic C. diffusa subsp. diffusa. The C. obliqua and C. rufipes complexes also need critical attention. The C. rufipes complex seems to be exclusively Neotropical, while the C. obliqua complex is here confirmed to be Pantropical, reaching Asia. These two species groups are historically problematic, and many names have been accommodated under one concept or another, depending on the author. It is very likely that both complexes will need to be studied concomitantly in order to fully understand their phylogenetic history, taxonomy and nomenclature.