Dioscorea nipensis Howard (1947: 119)

Raz, Lauren, 2016, Untangling the West Indian Dioscoreaceae: New combinations, lectotypification and synonymy, Phytotaxa 258 (1), pp. 26-48 : 40-41

2. Dioscorea nipensis Howard (1947: 119) . Type:— CUBA. Oriente (Holguin): Moa, 15 km. southwest of Compañia de Moa mill, 25 Jul 1941, Howard 5873 ♂ (holotype GH! [30362], isotypes GH! [30361], US!)

Dioscorea linearis Grisebach (1866: 251) . Dioscorea grisebachii Britton ex León (1946: 321) . Dioscorea raveni Ayala (1984: 296) . Type:— CUBA. Without locality, 1860–1864, Wright 3254 ♀ (holotype GOET!, isotypes G! (♀ and ♂, partial contents of the fragment packet, see below), S! (♀ part “B”, right hand side, designated here), BM! (♀ part “B”, designated here), GH! (♀ part “II”, right hand side, designated by Uline), MO! (♀ part “B”, left hand side, designated here), phototype NY! [of the G isotype, packet closed]).

Dioscorea montecristiana Hadač (1970: 430) . Type:— CUBA. Oriente (Guantanamo): jugo Montecristo dicto, solo calcáreo, 27 Jan.1968, Hada č 1322 ♀ (holotype PR!).

Rajania linearis auct. non (Griseb.) Uline ex R.Knuth in Knuth (1924: 168; see discussion below).

Notes: Endemic to eastern Cuba. The nomenclatural history of this taxon is particularly complex. The binomial Dioscorea linearis Bertero ex Colla (1836: 11) is the earliest publication of this name, referring to a species of Chilean origin (Bertero 1787, holotype TO) that had already been described three years earlier as D. saxatilis Poepp. ; Colla’s name was therefore reduced to a synonym. “ Dioscorea linearis ” reappeared in Grisebach (1866) with a brief description based on “Wright 3254” from Cuba. The type collection includes at least six duplicates (those cited here), of which five have mixed elements belonging to two different species with convergent leaf morphologies. Grisebach’s protologue does not specify which duplicate(s) he examined, but at the time of his preparation of the Catalogue he was based in Gottingen. The GOET sheet alone bears his handwritten diagnosis, as well as a label from the “Herbarium Grisebachianum”. The GOET sheet was also annotated as the Holotype by Richard Howard in 1984. At that time Howard was researching an article on the Cuban collections of Charles Wright (37 years after the publication of D. nipensis and R. linearis ). The GOET sheet is the only one of the six known duplicates that is 100% consistent with Grisebach’s description, which includes only characters of the leaves and capsules; there is no description of a staminate inflorescence, just the symbols “ ♂ …”. The GOET sheet is also the only duplicate that is not composed of two species. It bears a single element: a stem segment with leaves and infructescence attached. Yet, in his monograph of the Dioscoreaceae, Knuth (1924) cited the “typus” as Wright 3254 (♀) pro parte, from “herb. Berol.” There remains some uncertainty as to whether a duplicate was ever stored at B. As of September 2013, none could be found, but even if such a duplicate exists (or existed), it should not be considered the holotype.

The G sheet has an annotation label from the Museum Botanicum Berolinense but this cannot be considered evidence of the existence of a duplicate at B. The annotation was written by Uline who was at the time a doctoral student in Berlin (with access to the Museum’s annotation labels). Of the six known duplicates, only the S sheet bears an annotation by Knuth (no date, no “ typus ”). It consists of a staminate specimen of Dioscorea porulosa (part “A”, designated here, left hand side) and a fruiting specimen of Dioscorea nipensis on the right (part “B”). Yet the only specimen cited by Knuth (1924) as supporting material for his treatment of D. linearis Griseb. (p. 168), is the pistillate “ type ”. He made no reference to any staminate material, and yet he described a male inflorescence with pedicillate flowers bearing six fertile stamens, consistent with the flowers of D. porulosa . Aside from the S sheet, the only other staminate material from Wright 3254 is in the packet of the G sheet (not annotated by Knuth), which includes a small fragment each of a male inflorescence of D. porulosa and D. nipensis (the only male fragment of the latter species in any of the duplicates). The packet also contains loose leaves of both species and a few capsules of D. nipensis . Affixed to the G sheet is a pistillate specimen of D. porulosa (det. by Uline as R. microphylla Kunth ).

On the basis of the androecium with six stamens, Knuth assigned D. linearis Griseb. to Dioscorea sect. Apodostemon Uline. It was subsequently excluded from this section by Téllez and Geeta (2007). Since there is apparently no B sheet, it remains unclear what the source material was for Knuth’s description (1924) of D. linearis , or if any of it was adapted directly from Uline’s unpublished work (see below for discussion of GH sheet).

The BM sheet was determined by me in 2004 as R. porulosa , but at the time I failed to open the packet, which contains mature capsules of D. nipensis . There is also a very small piece of D. nipensis on the right side of the sheet: one leaf and a deteriorated pistillate inflorescence with sessile flowers. It is mixed in with what is mostly pistillate D. porulosa and is difficult to distinguish from the mass.

The remaining sheets require explanation because they are the base elements of three more names published after Knuth. In León’s Flora de Cuba (1946) the illegitimate name Dioscorea linearis Griseb. was replaced by the new name D. grisebachii Britton ex León , with no citation of a holotype or lectotype (only “Wright 3254”). Ayala (1984) pointed out that the León name was also illegitimate, because it had already been applied by Kunth (1850) to an unrelated Dioscorea species from Brazil. At this time, Ayala was based at MO and the only sheet of Wright 3254 that he examined was the MO sheet: it includes a pistillate Rajania element and a fragment of an infructescence consistent with the fruits of D. nipensis . Having never seen the GOET sheet, Ayala selected a new name for the Grisebach species, D. raveni , without recognizing the Rajania element, and designating the MO sheet (in its entirety) as the lectotype.

The GH sheet includes a pistillate Rajania element and a detached infructescence of Dioscorea nipensis . The latter was determined a century ago by Uline, as D. tamoidea Griseb. , but at the time he was unaware of the existence of two (as yet undescribed) endemic Dioscorea species from Eastern Cuba. Howard (1947), who was the first to acknowledge the mixed character of this Wright collection, recognized the Rajania element and published a formal description based on the GH sheet, giving it the new combination: R. linearis (Griseb.) R.A. Howard (see D. porulosa ).

There are three problems with Howard’s name: 1) Not having seen the holotype of D. linearis Griseb. (at the time he thought it was destroyed in WWII), and having accepted Uline’s determination of the capsules on the GH sheet, Howard was under the impression that Grisebach had mistakenly described the fruits of D. linearis from capsules of D. tamoidea , and the leaves from D. porulosa . Had he been aware of the D. linearis holotype in 1947, it is unlikely that he would have sunk the Grisebach name. 2) Howard was not the first to describe the Rajania element. Rajania porulosa was described in 1917 by Knuth (not based on Wright 3254); it corresponds morphologically and geographically to the Rajania element of the GH sheet. Howard never mentioned R. porulosa and it appears that he never saw any of the material cited by Knuth under this name. 3) Knuth (p. 168, 1924) cited “ Rajania linearis Uline in msc. p. p.” as a synonym of D. linearis , thereby establishing the name, even though he misapplied it. It seems that Uline’s intention (based on his annotation of the GH sheet), was to apply this name only to the Rajania element of Wright 3254 (GH). A half century later, Howard (1947) published the combination as his own: Rajania linearis (Griseb.) [Uline in msc. ex Knuth in syn.] Howard comb. nov..

Ironically, in the same 1947 paper, Howard also described a new Dioscorea species from Eastern Cuba, giving it the epithet D. nipensis . The species was described only from the type, collected by Howard, himself. This “new” Dioscorea had been documented by numerous collections in NY, US and HAC, from the early 1940s and earlier, but Howard did not cite any of this material, nor any of the Wright 3254 duplicates that contain a linear leaved Dioscorea element. His type is a staminate specimen, and he never described a pistillate plant because he hadn’t seen one, nor had he seen a large enough sample to interpret intraspecific variation. This species is highly variable in width and outline of the leaf base, but inflorescence morphology of both male and female specimens remains constant, independent of the variation in leaf outline. Howard had actually described the mate of the Wright 3254 GOET sheet without realizing it, and his is now the correct name for this species. Dioscorea raveni is superfluous, not because there is no species that corresponds to “ D. linearis Griseb. ” as argued by Téllez and Geeta (2007) who were also unaware of the other Wright duplicates, but because it was published 37 years after D. nipensis .

Finally, there are two duplicates of Wright 3254 at HAC, but both are unmixed sheets of D. porulosa , one with staminate inflorescences, the other sterile.