Recent phylogenomic analyses placed Albizia carbonaria Britton as the sister-group of the two currently known species of Pseudosamanea Harms, clearly outside Albizia section Arthrosamanea (Britton & Rose) Barneby & J.W. Grimes where it has until now been included. Its morphological similarities to Pseudosamanea are discussed, including characteristics of the polyad, and it is concluded that the species is compatible with the generic description of that genus except for its much more finely divided leaves with smaller leaflets, and its smaller flowers and fruits. Since these are merely quantitative differences, the species can readily be accommodated in Pseudosamanea. The new combination Pseudosamanea carbonaria (Britton) E.J.M. Koenen is made, and a diagnosis distinguishing it from the other two species of Pseudosamanea is presented.

Albizia sect. Arthrosamanea, polyads, Pseudosamanea, taxonomy

Recent phylogenomic analysis (Ringelberg et al. 2022) placed Albizia carbonaria Britton as the sister-group to Pseudosamenea Harms, separate from other Neotropical species of Albizia Durazz. that Barneby and Grimes (1996) placed in their section Arthrosamanea (Britton & Rose) Barneby & J.W. Grimes. Given the finely divided microphyllidious foliage and slender inflorescences of A. carbonaria, with much smaller flowers than those of the macrophyllidious species of Pseudosamanea, this may at first sight seem an unexpected phylogenetic relationship. However, on closer inspection, the similarities between A. carbonaria and Pseudosamanea are immediately apparent (Fig. 1) from the umbellate capitula with distinctly pedicellate peripheral flowers and a single enlarged sessile central flower (Fig. 1H), the ferrugineous indumentum on the twigs, leaf axes, peduncles, flowers and pods, the distinctive exfoliating bark (Fig. 1B, G), and the plano-compressed fruits and papery texture of the pod valves (Fig. 1D, I). In particular, the strongly dimorphic nature of the capitula, with only the much larger central flower being sessile, is unlike any of the species in Albizia sect. Arthrosamanea, but is effectively a miniature version of the capitula of Pseudosamanea guachapele (Kunth) Harms (Fig. 1C, F, H).

Figure 1. 

A–D Pseudosamanea carbonaria (Britton) E.J.M. Koenen A habit B trunk with exfoliating bark C inflorescences D pods E–I Pseudosamanea guachapele E habit F inflorescences G trunk with exfoliating bark H close-up of strongly heteromorphic inflorescence typical of the genus I pod A Bioexploradores Farallones B Karen Osorio C Juan Manuel de Roux D Juan Carlos Delgado Madrid E, H Colin Hughes F Cynthia Tercero G Bribrábrico I Daniel H Janzen A, B, C, D, F, G, I from https://www.gbif.org, distributed under a Creative Commons BY-NC-SA 3.0 License.

Albizia carbonaria clearly differs from Ps. guachapele and Pseudosamanea cubana (Britton & Rose) Barneby & J.W. Grimes in having more numerous and smaller leaflets (i.e. microphyllidious instead of macrophyllidious, Barneby and Grimes 1996), that are otherwise similar by being discolorous and in possessing a single midrib with pinnate secondary venation which also differs from the typically palmate or palmately-pinnate venation of leaflets of Albizia sect. Arthrosamanea. The stipitate, indehiscent or tardily dehiscent fruits of A. carbonaria are more similar to those of Ps. cubana, while those of Ps. guachapele dehisce actively along the ventral suture while the pods are still on the tree. Pods in all three species have the same characteristic ferrugineous indumentum. Thus, the flowers, leaflets and pods of A. carbonaria are merely quantitatively different from those of Pseudosamanea, being (much) smaller relative to Ps. guachapele and Ps cubana.

Based on its 32-celled polyads (Niezgoda and Nevling 1979a: figs 1 and 2; confirmed by my own observation, G.P. Lewis 3862), A. carbonaria was previously excluded from Albizia by Niezgoda and Nevling (1979a) who suggested it was preferable to restrict (Neotropical) Albizia to those species with 16-celled polyads. They placed A. carbonaria in Pithecellobium Mart., generating the only homotypic synonym of the species that I am aware of, which is somewhat remarkable for a species of the ingoid clade that was first described nearly a century ago, but all other authors (Nielsen 1985; Barneby and Grimes 1996; Rico-Arce et al. 2008) included it without question in Albizia. It is notable that Pithecellobium, under its current circumscription, is reported as having consistently 16-celled polyads, while Albizia sect. Arthrosamanea has either 16- or (28-)32-celled polyads (Guinet and Grimes 1997). Unfortunately, Guinet and Grimes (1997) described the characteristics of the pollen by genus, and failed to list which Albizia species have 32-celled polyads. However, Niezgoda and Nevling (1979a) compared the 32-celled polyads of Pithecellobium daulense Spruce ex Benth., which is (somewhat ironically) now a synonym of Albizia pistaciifolia (Willd.) Barneby & J.W. Grimes, with 16-celled polyads of Albizia retusa Benth. However, no voucher is cited for the P. daulense polyad in that publication, but in Niezgoda et al. (1983) the same species is included, with voucher Maguire 56150 which has since been re-identified as Enterolobium gummiferum J.F. Macbr. in NY (specimen consulted online, http://sweetgum.nybg.org/science/vh/, Barcode: 00924167). Further reason to doubt that A. pistaciifolia has 32-celled polyads is that in mimosoid species with polyads the fruit is usually pollinated by a single polyad (Niezgoda and Nevling 1979a; see also Banks et al. 2010), suggesting that species with 16-celled polyads should have a maximum of 16 seeds per fruit, and A. pistaciifolia is described as having 11–15 seeded fruits (Barneby and Grimes 1996). If A. pistaciifolia indeed has 16-celled polyads, it is possible that when A. carbonaria is excluded, Albizia sect. Arthrosamanea will be fundamentally characterized by 16-celled polyads. This seems likely for two reasons: (1) all taxa whose pollen has been studied in the genera Hydrochorea Barneby & J.W. Grimes (sensu Soares et al. 2022), Jupunba Britton & Rose and Punjuba Britton & Rose (i.e. the “Abarema alliance” of Barneby and Grimes 1996), that together form the sister-group of Albizia sect. Arthrosamanea in the Jupunba clade (sensu Koenen et al. 2020; Ringelberg et al. 2022), consistently have 16-celled polyads (Guinet and Grimes 1997); (2) all species included by Barneby and Grimes (1996) in Albizia sect. Arthrosamanea have fruits that are usually c. 11–15 seeded (although occasionally up to 18-seeded in Albizia inundata (Mart.) Barneby & J.W. Grimes, which accordingly appears to have 16(-20?)-celled polyads, see Vossler 2019: fig. 2c & e), except for A. carbonaria that has 20–26-seeded fruits (Barneby and Grimes 1996).

Pseudosamanea guachapele has 32-celled polyads (Niezgoda and Nevling 1979b; Guinet and Grimes 1997); therefore those of A. carbonaria are compatible with Pseudosamanea. The polyads of Ps. cubana are reported to be 24-celled (Guinet and Grimes 1997), but the fruit is described as 24–30 seeded (Barneby and Grimes 1996), casting doubt on the reported number of grains per polyad, or suggesting that the species is perhaps variable in grain number per polyad. Given the variation in the number of cells per polyad within several genera as well as within species or even individuals (e.g. Leucochloron Barneby & J.W. Grimes; Guinet and Grimes 1997), this character has been viewed as unimportant for generic delimitation in recent decades, despite mention by Guinet and Grimes (1997) that several groups are invariant in this regard. Phylogenetic studies are now revealing that polyad grain number may be more useful than previously thought. For example, the Calliandra clade (sensu Koenen et al. 2020) comprising Acaciella Britton & Rose, Calliandra Benth. and Afrocalliandra E.R. Souza & L.P. Queiroz has consistently 7 or 8-celled polyads, which is unusual in the ingoid clade where species usually have ≥ 16 cells per polyad. Similarly, I suggest that the entire Jupunba clade sensu Koenen et al. (2020) may be fundamentally 16-celled, and many of the genera that Guinet and Grimes (1997) described as having > 16 cells per polyad or of variable polyad number are placed in the Inga clade of Koenen et al. (2020; see also Ringelberg et al. 2022). All this suggests that the utility of polyad grain number should be re-evaluated in the light of recent phylogenomic evidence. In conclusion, the decision of Niezgoda and Nevling (1979a) to exclude A. carbonaria from Albizia based on the grain number per polyad was correct and in line with the other morphological differences discussed here; however, they did not place it in the correct genus. Instead, the 32-celled polyads provide further evidence in support of placing the species in Pseudosamanea.

An important character used by Barneby and Grimes (1996) to define Pseudosamanea is inflorescences axillary to coeval leaves (although in Ps. guachapele flowers often develop before the leaves). The inflorescences of A. carbonaria appear somewhat similar to those typically found in Albizia sect. Arthrosamanea and many other mimosoid genera where the capitula form a compound inflorescence resembling a pseudo-raceme or a complex panicle of pseudo-racemes with suppression or early shedding of leaves. However, it is questionable whether the description of the inflorescences of Pseudosamanea by Barneby and Grimes (1996) is completely accurate, as specimens at K (S.C. Sant’Ana 1023 and C.E. Hughes 753) have short compound pseudo-racemose inflorescences, comprising multiple umbellate capitula arising from a leaf axil (rather than only singly as described for Ps. guachapele by Barneby and Grimes 1996), with the subtending leaves being caducous. Another key feature of the inflorescences of Pseudosamanea is that the meristem at the apex of reproductive branches continues to grow (i.e. is indeterminate), producing new leaves below which the fruits develop. Unlike in other species of Neotropical Albizia, this is clearly also the case in A. carbonaria, see e.g. E. Suclli & J. Farfán 1258 – NY, Barcode 1300235; J. Leon 4372 – F, https://collections-botany.fieldmuseum.org/, Catalog number 1578909), providing evidence that the inflorescences of A. carbonaria are not qualitatively different from Pseudosamanea, while being unlike the efoliate pseudoracemes typically found in Albizia sect. Arthrosamanea where the meristem is not continuous.

Barneby and Grimes (1996: 226) noted that “among related albizias with multifoliolate leaves, A. carbonaria is notable for the dorsally pallid leaflets, the dense golden indumentum of the inflorescence, the distinctly pediceled flowers, and the papery tomentulose pod.” All these characters distinguishing it from other Albizia sect. Arthrosamanea, are in agreement with the generic description of Pseudosamanea (sensu Barneby and Grimes 1996). Albizia carbonaria differs from Pseudosamanea as delimited by Barneby and Grimes (1996) in the dimensions and number of leaflets per leaf, flower dimensions and inflorescence structure, but these are minor quantitative differences that do not constitute sufficient reason for not including the species in Pseudosamanea, an option which would require segregation of a new monotypic genus. I therefore propose to include the species in Pseudosamanea and hence the new combination is made:

I wish to thank Gwil Lewis for acting as my SYNTHESYS+ host which made for an excellent visit to the legume collections at K and Colin Hughes, Luciano Paganucci de Queiroz, Stefanie Ickert-Bond and Ethiene Guerra for comments and suggestions on the manuscript that have greatly improved the text. Thanks to the photographers for making their images available for the figure in this article. My research is supported by the Swiss National Science Foundation (Early.Postdoc.Mobility fellowship P2ZHP3_199693 to EJMK). This research received support from the SYNTHESYS+ project http://www.synthesys.info/ which is financed by European Community Research Infrastructure Action under the H2020 Integrating Activities Programme, Project number 823827.