﻿Notes on Brazilian Pachira (Malvaceae, Bombacoideae) II: Additional typifications and new synonymies

﻿Abstract The typification and status of the names of 14 species of Pachira (Malvaceae: Bombacoideae) found in Brazil are discussed, including type material from Brazil, the Guianas, Colombia, Venezuela, and cultivated in Algeria. We designate 11 lectotypes, three neotypes, and four epitypes for these names. Six names are newly considered to be synonyms of the species accepted here. The results support a forthcoming taxonomic treatment of Pachira for the Flora of Brazil.


Introduction
Pachira Aubl. is the most species-rich genus among the 17 genera of Bombacoideae (Malvaceae) (Carvalho-Sobrinho et al. 2016) and it consists of trees distributed primarily in wet forest in northern South America (Robyns 1963;Alverson 1994;Fernández-Alonso 1998, 2003Carvalho-Sobrinho et al. 2014). About one-hundred names have been published in Pachira (IPNI 2021; Tropicos 2021), but they probably represent only about 50 species in this genus. The genus was last revised by Robyns (1963). Since then, generic concepts have changed radically and a contemporary revision of Pachira would resolve not only questions about species richness, but also clarify evolutionary relationships and biogeography.
Herein we discuss the typification and status of the names (and some of their synonyms) of 14 species of Pachira found in Brazil. Type material is from Brazil (including Brazilian material cultivated in Algeria), the Guianas, Colombia, and Venezuela. We designate 11 lectotypes, three neotypes, and four epitypes for these names. Notably, an epitype is designated here for P. aquatica Aubl., the type of the genus. Six names are newly considered to be synonyms of the species accepted here. Most significantly, P. nitida Kunth, a name often used in checklists of the Brazilian flora (see e.g., Duarte 2010;BFG 2015), is revealed to be a synonym of P. minor (Sims) Hemsl. Similarly, P. dolichocalyx A. Robyns, previously considered to be endemic to French Guiana, is shown to be a synonym of P. macrocalyx (Ducke) Fern. Alonso that was described from Brazil.

Materials and methods
Protologues of names of Pachira taxa found in Brazil (and their basionyms) were examined along with relevant revisionary and floristic literature in order to determine what constituted original material, the identities of these taxa, and to establish whether these names had been typified. The specimens cited as types were either examined by us in person or via digital proxies. Herbarium acronyms for these specimens follow Thiers (2021). The notation "F neg. no." refers to the "Berlin Negatives" of the Field Museum (F), a unique type-photographic collection of European herbaria that J. Francis Macbride began assembling in 1929 (https://www.fieldmuseum.org/node/5186). These photographic images frequently are the only records of types or original material that was destroyed during WWII and these images are commonly found in northern hemisphere herbaria.
For the sake of brevity, we do not provide complete synonymies for the Pachira taxa that we discuss, but rather list only names that have not been, or were previously incorrectly, typified. In a few instances, we include names that were typified earlier, but whose citations require clarification. We also list names of taxa that are considered here to be new synonyms irrespective of their typification status. Note. Yoshikawa et al. (in press) typified several names that are synonyms of Pachira aquatica, but omitted mention of nomina nuda. Despite their lack of nomenclatural standing, these nomina nuda associated with P. aquatica do appear in standard indices (e.g., IPNI 2021; Tropicos 2021) and we include them to clarify their identification. Likewise, we discuss two names typified by Yoshikawa et al. (in press) in order to clarify their synonymies and bibliographic citations that are confused in standard indices.
The lectotype of Pachira aquatica consists solely of detached leaflets and a sterile twig. Consequently, a flowering collection from French Guiana is designated here as an epitype (Fig. 1).
"Carolinea pompalis fl. mex. ic. ined." was cited originally as a synonym of C. minor Sims (≡ Pachira minor (Sims) Hemsl.). Robyns (1963) did not agree with this interpretation and placed the nomen nudum among his "Species dubiae incertae sedis." McVaugh (2000, p. 88) discussed the sources of the material that de Candolle (1824) examined, which included a plate in G (see F neg. no. 30513) and presumably another illustration now in the Torner Collection (Hunt Institute for Botanical Documentation acces-  sion No. 6331.0864; see also No. 6331.1977). These images depict P. aquatica, which occurs in Mexico and South America. Pachira minor is restricted to South America, which was not visited by the Sessé and Mociño Expedition.
Carolinea macrocarpa was described from Mexican material cultivated in Berlin. The protologue gives the type locality as "Ad ripas fluminum et rivulorum Papantlensium et Tecolutensium. Papantlae." Robyns (1963, p. 239) selected "Schiede et Deppe 1312" as "typus" and he cited specimens at BM, LE and W, but not HAL. The labels on the HAL specimens, however, suggest the collecting combination may have been "Deppe & Schiede" and not "Schiede & Deppe." Pachira longifolia is a superfluous name for P. macrocarpa (≡ Carolinea macrocarpa). The epithet of this superfluous name is often miscited as "longiflora" (see e.g., Robyns 1963;Alverson in Berry et al. 1997), which is perhaps understandable given that Hooker (1850) Robyns (1963, p. 203) designated "Riedel s.n. (LE)" as lectotype of Bombax calophyllum. When Fernández-Alonso (1998) made the combination in Pachira, he accepted Robyn's type designation, but stated that he had not seen the lectotype. Robyns annotated the two Riedel specimens in St. Petersburg (LE) cited above as "Bombacopsis calophylla (K. Schum.) A. Robyns, comb. nov." and wrote "lectotypus" on the one specimen that has a handwritten label indicating the type locality as "Brasiliae: R. Janeiro," locality information that matches the protologue. The ICN (Turland et al. 2018;Art. 7.10) requires that a type designation be effectively published and the mere annotation of a herbarium sheet does not meet this requirement. Our lectotypification (second step; see Turland et al. 2018;Art. 9.17) narrows the type to a single element, a gathering that has loose leaves and a fragmented flower (calyx, ovary, and androecium). The isolectotype is sterile.
The protologue of Bombax stenopodum cites a single collection, "E. Ule n. 4631," which was deposited in Berlin (B †; see F neg. no. 9545). Inasmuch as the Berlin specimen was destroyed in WWII and no duplicate material has been found, we designate a neotype for this name. Our choice makes this name an obligate synonym of Pachira calophylla.  Robyns (1963) described Rhodognaphalopsis duckei and R. d. var. obtusifolia he had very little material available. In fact, he cited fewer than a dozen collections for both taxa. The characters that he used to distinguish var. obtusifolia from the nominate variety were subtle and overlap with specimens he cited as paratypes of the latter. We cannot distinguish the obtuse leaflet bases of Ducke s.n. [RB 23484] (see e.g., K barcode 000913930), a paratype of R. duckei, from those of Ule 6077 (see e.g., K barcode 000382340), the type of R. d. var. obtusifolia. For this reason and because the two taxa lack discrete ranges, we propose these names should be treated as synonyms.  (1831), which is not original material and thus his designation can be superseded (Turland et al. 2018;Art. 9.19(a)). The published Vellozo plates were prepared and issued posthumously and there is no evidence that Vellozo ever saw them. Original pen and ink drawings that serve as the basis for these published plates, however, are archived in the Biblioteca Nacional do Brasil, Rio de Janeiro (see http://objdigital.bn.br/acervo_digital/div_manuscritos/ mss1198656/mss1198656.htm).
Delprete et al. (2019) re-examined the typification of Pachira stenopetala and determined that the holotype cited by Carvalho-Sobrinho et al. (2013) did not agree with the protologue and they further established that the only original material available is the specimen in the Casaretto Herbarium (TO) that they designated as lectotype.
The specimen designated here as lectotype of Pachira elegans is labeled "Pachira? elegans, Daniel" in Planchon's hand. It was collected in 1854 before the name was published and it is now in the Herbier Planchon (MPU). It therefore can be considered original material (see Turland et al. 2018;Art. 9.4(a)) even though it is labeled "Hort. Paris" and not Hortus Donatensis. In addition, although the specimen is sterile the description of setiform mucros terminating leaflet apices matches the physical specimen. The protologue also cites "Brésil. Rio de Janeiro, Gaudich. in Herb. Mus. Par.," which almost certainly is a reference to two collections made by Ch. Gaudichaud that have leaves that are superficially similar to Pachira endecaphylla. These collections (Gaudichaud 945 and 945bis) were discussed in detail by Carvalho-Sobrinho et al.   Robyns (1960) established that there is no extant original material of Pachira glabra. He then neotypified (first step) the names P. glabra and P. oleaginea with the same collection (i.e., "Rivière s.n."). This collection is represented by three sheets in Paris (P), all three of which were labeled "Neotypus" by Robyns. The three sheets are not marked as being part of the same specimen and we interpret them as duplicates of a single gathering (see Turland et al. 2018;Art. 8.3). Accordingly, we designate here one of the three sheets as the neotype (second step) for the names P. glabra and P. oleaginea. One of the isoneotypes (P barcode 02286301) has a faint pencil sketch of the fruit that was labeled by Decaisne "fructus maturus magnif. nat." The labels on the neotype and one of the two isoneotypes indicate that the collection was made by "Cl. Ch. Rivière" where "Cl." is Latin for clarissimus (i.e., distinguished). The latter label also has the initial "A." added below this name, presumably a reference to A. Rivière who, however, was deceased before the collection was made. The horticulturists and brothers [Marie] Auguste Rivière (1821-1877) and Charles Marie Rivière (b. 1845) were successive directors of the Jardin du Hamma, the latter succeeding the former (Stafleu and Cowan 1983).  (1881) gives no explicit indication as to where he examined material. The description of this species (and others in the same article) ends with the phrase "H. Mus. Par." Presumably this is a reference to material in the herbarium and not the gardens ("hortus") in Paris since Ch. Rivière, Director of the Jardin du Hamma in Algeria is acknowledged by Decaisne for providing him with fresh flowers and fruit of Pachira species.

Pachira humilis
The synonymy and authorship of Pachira humilis Spruce ex Decne. have been confused because of the mistaken belief that Bentham (1862) transferred the species to Bombax. Bentham cited the unpublished name "Pachira humilis, Spruce" in a discussion regarding the distinction between Bombax and Pachira, but he did not definitely associate the final epithet "humilis" with Bombax (see Turland et al. 2018;Art. 35.2). Schumann (1886, p. 224) was the first to make the combination in Bombax, albeit inadvertently. He recognized "Bombax humile, Benth." and he cited the basionym when he wrote "Pachira humilis Spruce Msc.! in schedulis; Dcne. Miscell. bot. 1880, p. 10." The subordinate phrase is a reference to "Miscellanea botanica," which reprints Decaisne's (1881) article on Bombax and Pachira including the original description of P. humilis.  (1881) cites "Carolinea insignis, Lodd., Bot. Cab., 1004, (non Swarts)." The "name" ascribed to Loddiges, however, has no standing. Loddiges (1825) simply illustrated C. insignis Sw. without explicitly stating that he was illustrating the species previously described by Swartz. Robyns (1963, p. 250) designated a lectotype (as "holotypus") for P. loddigesii, but this was incorrect since the name is typified automatically by the type of the name which ought to have been adopted under the rules (Turland et al. 2018;Art. 7.5). Similarly, Turner (2016Turner ( , p. 1115 argued that "Carolinea insignis G. Lodd." is a valid name, even though he noted that Loddiges cited Swartz in his description and thus provided an indirect reference to C. insignis Sw. Turner's lectotypification of this "name" incorrectly attributed to Loddiges also is unnecessary.

Pachira insignis (Sw.) Sw. ex Savigny, in
A lectotype (second step) is designated here for the name Carolinea affinis because although Robyns (1963, p. 251) stated the "holotypus" was in Munich (M) there are two sheets in that herbarium. Robyns annotated both sheets as Pachira insignis, but did not indicate that either was type material. The specimen with label data that matches the type locality given in the protologue is designated here as the lectotype.
We also designate here a lectotype for Pachira spruceana. Neither Ducke (1925) nor Robyns (1963) selected one for this name. We have chosen a specimen deposited at Paris (P) that has handwriting on Spruce's label that matches the type locality and collecting date ("Prope Panure ad rio Uaupes, Oct 1852-Jan 1853") cited in the protologue.
Bombax spectabile was proposed as a replacement name for B. insigne (Sw.) K. Schum., which is a later homonym of B. insigne Wall. Bombax insigne Wall. is a Paleotropical species found in India, south-central China, and south-east Asia. Note. When Robyns (1988) described Pachira dolichocalyx he compared it to P. aquatica and P. insignis, which according to his concept of the Bombacoideae were the only two species that comprised Pachira. We assume he failed to compare the material to Bombacopsis macrocalyx because he considered Bombacopsis to be a different genus.
In the protologue of Pachira dolichocalyx, the petals are described as greenish outside and reddish-purple inside "in vivo." This information could only have come from whatever may have been inferred from examining the pressed and dried material and information on the type label, which simply states that the petals are wine red ("rouges lie de vin"). Robyns (1963, p. 204) described the petals of P. macrocalyx (as Bombacopsis macrocalyx) as pale yellow outside and whitish-puberulent inside. Yet many specimen labels of P. macrocalyx report red petals. Additionally, staminal-tube length, fruit and seed morphology (mainly dimensions), and habitat are similar for both species supporting our decision to consider them synonymous. ( Note. In the protologue of Carolinea minor, Sims (1811) states that he received this plant from "Messrs. Loddiges and Sons, under the name of Bombax Carolinoides, an appellation given it by Dr. Anderson of the Botanic Garden at St. Vincent's." Thus, original material that can be considered for a lectotype is either a cultivated specimen (or specimens) from Loddiges nursery in Hackney (now London) or the plate (t. 1412) illustrating Carolinea minor. Robyns (1963, p. 280) stated that the "holotypus" was at BM ("Anderson (321 ?) (BM)"), but while it is clear that Anderson conveyed seed to Loddiges, Sims's description was not based on Anderson's wild-collected herbarium material and such specimens are not original material. Robyn's "holotypus" cannot therefore be corrected to neotype (see Turland et al. 2018;Art. 9.10). Robyns (1963, p. 284) effectively designated a lectotype for the name Pachira nitida when he stated that a collection from Caño de Pimichin made by F.W.H.A. von Humboldt and A.J.A. Bonpland was the "holotypus." The handwriting on the label of this specimen (P barcode 00679764) is that of Kunth (see Stauffer et al. 2012) and the fragmentary nature of the flower agrees with statements in the protologue (here translated) that declare "Only fragments available to us. A description of a flower from the label of Bonpland." The handwriting on the label of the isolectotype (P barcode 04694524) is that of Bonpland (see Stauffer et al. 2012). This specimen was originally identified as Carolinea princeps L.f. (≡ P. aquatica), which also is noted in the protologue where P. aquatica is listed as a synonym of P. nitida with doubt. The morphology of the leaves (leaflet shape, venation, and strongly raised midribs on abaxial surfaces) and of the flower (calyx shape and dimensions and stamen length) of the type of P. nitida matches that of the type of P. minor. Note. Robyns (1963, p. 201) designated "For. Bur. 1901" as the lectotype of Bombax nervosum. He annotated the specimen as "Bombacopsis nervosa (Uitt.) A. Robyns, comb. nov." and as "lectotypus!" Fernández-Alonso (1998, p. 310) stated that he had not seen the lectotype when he transferred Bombax nervosum to Pachira. The lectotype of B. nervosum, however, is sterile and to avoid any ambiguity about its identity, an epitype is herein selected. The epitype has flowers, a determination label in Robyns's hand, and it was cited in his revision (Robyns 1963, p. 201). Bombax obtusum (Spruce ex K. Schum.) Bakh., Bull. Jard. Bot. Buitenzorg, sér. 3, 6: 181. 1924. Bombax poissonianum K. Schum., in Martius, Fl. Bras. 12 (3) Note. The protologue of Pachira obtusa cites a single collection, "Spruce n. 2150," which we assume Schumann examined in Berlin (B) and which was subsequently destroyed in WWII. In his revision, Robyns (1963, p. 284) failed to select a lectotype since he made no distinction between duplicates of this collection deposited in various herbaria. We designate here as lectotype of the name a sheet deposited in Kew (K) that has a handwritten locality as well as a narrower collecting date ("Feb. 1852") than the majority of duplicates that have printed labels and an eight-month range for the collecting date ("Jan.-Aug. 1852").

Pachira obtusa
In the protologue of Bombax poissonianum, Schumann (1886) states that he thinks his new species might be the same as the Pachira nitida of Decaisne or Kunth (viz., "Pachira nitida Dcne. ! Miscell bot. 1880. p. 9, an Kunth?"). Bombax poissonianum, however, is not a superfluous renaming of P. nitida because Schumann (1886, p. 225) included an expression of doubt (see Turland et al. 2018; Art. 52.2, Note 1). There is nothing to tell us who collected the type specimen nor when it was collected. The epithet and author (viz., "nitida Kth.") written on the holotype label appears to be a later addition (the handwriting differs from that of whomever wrote the locality). The specimen was annotated "det. Schumann in Fl. Bras." and is undoubtedly the one Schumann (1886, p. 225) mentioned in the protologue.
In the protologue of Bombax utiarityi, Hoehne (1927) cites a single collection, "Kuhlmann 2145," but he does not indicate where it was deposited. When Toledo and Handro in Hoehne (1944, p. 39) proposed a new combination for this species, they mentioned a Kuhlmann specimen housed at SP (as "Inst. Bot. 11.914, leg. J.G. Kuhlmann"), but they did not provide a collection number and failed to use the word "type" or an equivalent (Turland et al. 2018;Art. 7.11). When Robyns (1963, p. 285) proposed a new combination and status for B. utiarityi, he also effectively selected a lectotype (Turland et al. 2018;Art. 9.10). The lectotype he designated has an original handwritten label. Inasmuch as Bombax utiarityi agrees with both B. poissonianum and Pachira obtusa morphologically and the habitat of occurrence (flooded forest in Amazonia and adjacent Savanna) of all three are the same, we consider B. utiarityi to be a synonym of P. obtusa.
The designation Pachira utiarityi (Hoehne) Hoehne (Hoehne 1951) is not validly published. It was proposed as an alternative for Bombax utiarityi, which was the name Hoehne accepted (see Turland et al. 2018;Art. 36.1(a)). ( Robyns (1963, p. 207) stated that a Martius specimen in Munich (M) was the "holotypus" of the name Bombax retusum, he effectively selected a lectotype (Turland et al. 2018;Art. 9.10). However, there are at least two specimens with identical label data in that herbarium that must be considered syntypes as well as a copy of the plate cited in the protologue ("tab. 60," sphalm. pro 59), but published a year later (Martius 1826). Robyns wrote "holotypus" on one of the two specimens and annotated all three elements as "Bombacopsis retusa (Mart. et Zucc.) A. Robyns, comb. nov." The ICN (Turland et al. 2018;Art. 7.10) requires that a type designation be effectively published and the mere annotation of a herbarium sheet does not meet this requirement. Our lectotypification (second step; see Turland et al. 2018;Art. 9.17) narrows Robyns' selection to the specimen with the best flowering material. Robyns (1963, p. 205) attributed the basionym to "Mart. et Zucc.," but according to Stafleu and Cowan (1981, p. 329) authorship is to be attributed to von Martius alone. Fernández-Alonso (2003, p. 36) also included J.G. Zuccarini as a co-author of this species name. Interestingly, Schumann (1886, p. 226) recognized "Bombax retusum Mart." while citing "Bombax retusum Mart. et Zucc.! Nov. gen. et spec. I. 92. t. 59." The article that includes the validating description of B. retusum (Martius 1825)

Note.
A single collection, "R.E. Schultes & F. López 9342," is cited in the protologue of Bombax sordidum. When Robyns (1963, p. 292) proposed a new combination and status for this name, he indicated that the holotype was deposited in the Gray Herbarium: "Colombie: Vaupes: Schultes & F. López, 9342 (f., fr., holotypus GH)". There are, however, three sheets of this collection in that herbarium, and although one was annotated as the holotype by Robyns, the ICN (Turland et al. 2018;Art. 7.10) requires that a type designation be effectively published. The mere annotation of a herbarium sheet does not meet this requirement, hence our lectotypification (second step). We designate here the specimen that has the greatest abundance of indumentum on the abaxial surfaces of the leaflets because this character is considered diagnostic by Robyns (1963, p. 292, viz. "a specie foliolorum lamina infra densissime et sordide pulveraceolepidota sat differt").