Taxonomic revision of Myrosma ( Marantaceae )

Myrosma is a genus of rosulate herbs characterized by a compact and strongly monosymmetric inflorescence with conspicuous, white to pale green, mostly membranous bracts, in the axil of which is a single flower pair. The genus is widely distributed throughout Central Brazil, northern South America, and the Caribbean region. It occurs mostly in savannah environments but can also be found in humid and shaded habitats. We present a taxonomic revision of Myrosma recognizing a single species, Myrosma cannifolia L.f. A complete description of that species is provided with notes on its ecology, distribution, and variation. Three species and one variety are reduced into synonymy and four names are lectotypified. A complete nomenclatural account for all combinations including the name Myrosma is given.


INTRODUCTION
Myrosma L.f. (Marantaceae) is a monotypic genus distributed throughout the Caribbean region, the Guianas, Colombia, Trinidad, Venezuela, Brazil, Peru, and Bolivia.It has traditionally been considered closely related to Saranthe (Regel & Körn.)Eichler and other genera of Andersson's (1981Andersson's ( , 1998) Myrosma group (Hylaeanthe A.M.E.Jonker & Jonker, Ctenanthe Eichler, and Stromanthe Sond.) with which it shares characters such as flower groups with slightly to markedly elongated axes (subbrachyblastic to dolichoblastic in the terminology of Andersson 1998), lack of interphylls, and a short corolla tube.Phylogenetic studies based on molecular data (Andersson & Chase 2001, Prince & Kress 2006, Suksathan et al. 2009) have shown that these genera form part of a strongly supported monophyletic group (the Maranta clade) that also includes Maranta L., Monophyllanthe K.Schum., and Koernickanthe L.Andersson, as well as two more distantly related Asian-African genera, Halopegia K.Schum.and Indianthus Suksathan & Borchs.The precise generic relationships within this large group remain, however, unresolved.Andersson & Chase (2001) found no resolution for the position of Myrosma within the Maranta clade based on rps16 data, but they found strong support for a monophyletic Saranthe.When morphological data were added to the analysis a sister relationship between Myrosma and Saranthe was indicated.Prince & Kress (2006) did not include Myrosma in their analysis.
In this paper we present a modern taxonomic revision of Myrosma.We agree with Andersson (1998) in recognizing only a single species.We present a formal nomenclatural account for the genus and the species.Four names are lectotypified and four new synonyms are formally made.We also provide a complete morphological description of M. cannifolia including new information on its ecology and distribution documented by specimen records.Finally a nomenclatural account is given for all combinations involving the name Myrosma.

DISCUSSION
Our generic circumscription agrees with Andersson's (1981Andersson's ( , 1998) ) who considered Myrosma in a strict sense a highly distinct and characteristic genus, diagnosed by the combination of monosymmetrical, compact inflorescences with whitish green bracts, a large, petaloid staminal appendage, solitary flower pairs, and absence of bracteoles.Some species of Saranthe (notably S. unilateralis (Baker) L.Andersson) approach Myrosma in inflorescence morphology and the only floral character that seems to consistently distinguish the two genera is the size of the staminal appendage (smaller in Saranthe).Molecular data, however, do not indicate any close relationship between the two genera (e.g., Andersson & Chase 2001).Ongoing analyses of matK and rps16 intron data (Vieira et al. unpubl. data) suggest that Saranthe forms a well-supported monophyletic entity sister to Ctenanthe and Stromanthe, while Myrosma and Hylaeanthe form a distinct clade more closely related to Maranta.For this reason we prefer to maintain Myrosma as a separate genus, at least until phylogenetic relationships have been satisfactorily settled.Hylaeanthe differs from Myrosma in having the leaf sheath margin splitting up in a reticulum of fibers (fenestrate sheaths).
Habitat & Ecology -Mostly confined to savannah vegetation where it usually grows near water courses.An exception occurs in Bolivia where the species is found in humid and shady environments.Notes -Myrosma cannifolia most resembles Saranthe unilateralis with which it shares the characters of strongly monosymmetric and compact inflorescence, and white to pale green, persistent bracts.However, M. cannifolia has a larger staminal appendage (6 by 3 mm vs 4 by 1.5 mm in S. unilater alis), smaller leaf blades (11-20(-31) by 1.7-8 cm vs 33.5 by 12 -15 cm long), unequal outer staminodes (equal in S. uni lateralis) and larger flowers (1-1.5 cm vs 0.8 -1 cm long).Linnaeus f. ( 1782) mentioned only a single collection in his protologue of Myrosma cannifolia (C.G.Dahlberg 121).The specimen is found at LINN, being in perfect condition and agreeing completely with the original description.The type specimen is not numbered, but there is no doubt that this specimen is the holotype of the name and that the correct number of the collection is 121.Dahlberg was a Swedish soldier who collected for Linnaeus and who made in Suriname a collection of 186 plant specimens, which were first given to King Gustav III of Sweden and consequently, in 1774, transferred to Linnaeus (Stafleu & Cowan 1976: 588).During his fieldwork in Suriname Dahlberg made accurate field annotations, which were written down in his 'Catalogus der Vlessen, van de Boom, Struik, Plant en Rankgewassen, dewelke ik, in Spiritu vini bewaard heb'.This was a catalogue of all his collections, many of which were collected in spirit and which were later dried.Number 121 gives an extensive description in Dutch of the plant that is the type of M. cannifolia.The description starts (translated) with: "This is the leaf, inflorescence, and root of a plantlet called Turara, the flowers are white, sprouting laterally from the side of the stalk …".The Dutch text by Dahlberg was many years later translated into English by J.E.Smith (1819) "Gathered by Dalberg in Surinam.The root is creeping, with long hairy fibres …" and at the end "This was one of the plants which made a part of the Surinam collection, preserved in spirits, presented to Linnaeus by King Gustavus".Körnicke (1862) described Maranta cuyabensis (≡ Myrosma cuyabensis (Körn.)K.Schum.)based on two specimens from Central Brazil without a clear indication of a holotype.As diagnostic characters he cited narrowly elliptic leaf blades with narrower base and the glabrous or hirsute inflorescence peduncles.Indeed a number of specimens collected from cerrado and gallery forests from the Brazilian states of Goiás (e.g., H.S. Irwin et al. 34782 (NY)), Mato Grosso (e.g., G. Hatchbach 36149 (BR, C, GB, MBM, NY)) and southern Tocantins (e.g.,G. Hatschbach et al. 56022 (C,GB,MBM,NY)) as well as the eastern part of the Department of Santa Cruz in Bolivia (e.g., W.W. Thomas et al. 5687 (GB, NY)) do have extremely narrow leaf blades.However, as no other characters distinguish these plants from typical M. cannifolia, we have chosen not to recognize this variation as a formal taxon.As lectotype of Maranta cuyabensis we have chosen Riedel 857, as this collection is more complete than the other syntype (A.L.P. da Silva Manso & J. Lhotsky 85).The original material studied by Körnicke was most likely located at BONN but destroyed in 1941.As lectotype we have instead chosen the duplicate at G. Maranta moritziana was described by Körnicke in the same work as M. cuyabensis.It was transferred to Saranthe by Eichler (1884) and this decision was accepted by Petersen (1890).Schumann (1902) placed it in synonymy of Myrosma cannifolia.Study of the original description and the illustration included in Flora Brasiliensis corroborate that decision.Of the four syntypes listed by Körnicke (1862) two have been located (N.Funck 71 and M.R. Schombugk 1305).N. Funck 71 (P) has been selected here as lectotype.
Myrosma boliviana was described by Loesener (1915) from a Bolivian collection E.H.G.Ule 9247 (holotype B †).A duplicate of this collection is found at MO and has been chosen as lecto-type.The diagnostic character for the species was that it had elliptic to broadly ovate leaves.In the same article Loesener also described a new variety of his new species, M. boliviana var.acreana, based on a second collection from the same area as the holotype (E.H.G.Ule 9246) representing a plant with smaller dimensions.As noted above M. cannifolia displays a large and continuous variation in size and leaf shape and distinct taxa cannot be separated on that basis.However, a small number of specimens collected in the sub-Andean region of the western Amazon from Bolivia to Peru (e.g., T. Plowman 5998 (F), R. Vásquez 3877 (MO), M. Hagberg & E.-L.Medin 357 (GB)) tend to have one elongated internode basally on the flowering stem resulting in a clearly visible node some distance below the inflorescence, a character that we have not observed in any specimen from other parts of the range of M. cannifolia.The sub-Andean plants also have inflorescence bracts that are more fibrous and greenish brown (when dried) than those from other areas.If these characters prove to be consistent when more material becomes available it might be considered whether these specimens should be recognized as a distinct taxon.
Finally, study of the type specimen of Saranthe marcgravii, made from cultivated plants with origin in Central Brazil, as well as the original description and the excellent illustration provided by Pickel (1939) leave no doubt that this is identical to M. cannifolia.