The rise of Cynometra (Leguminosae) and the fall of Maniltoa: a generic re-circumscription and the addition of 4 new species

Abstract Cynometra L. is a genus of ca. 85 species of shrubs to large trees. It is amongst the largest genera in the legume subfamily Detarioideae and one of the few with a pantropical distribution. Perhaps due to this wide distribution and high diversity, systematists and taxonomists have struggled with the classification of Cynometra and its close ally, the genus Maniltoa Scheff. Recent phylogenetic studies have shown that many of the African species are more closely related to other genera and that the genus Maniltoa is nested within a clade of Indo-Pacific Cynometra. Here, I present an emended circumscription of Cynometra that excludes the African species defined by jointed pedicels and dehiscent fruits and includes the species formerly recognised in Maniltoa. New combinations in Cynometra are also provided for those species that require them. Additionally, four new species of Neotropical Cynometra are described and illustrated: Cynometracerebriformissp. nov. from the lower Rio Trombetas in Brazil; Cynometradwyeriisp. nov. from the Darien gap region of Panama; C.tumbesianasp. nov. from the dry tropical forests of Ecuador and Peru; and C.steyermarkiisp. nov. from the foothills of the western Cordillera de la Costa in Venezuela.


Introduction
The genus Cynometra L. (Leguminosae) has a broad tropical distribution, is relatively species rich and many of its species are poorly represented in herbaria. The approximately 85 species of trees and (some) shrubs in the genus are spread somewhat evenly amongst four regions (Figure 1): the Neotropics (from southern Mexico and the Caribbean to northern Argentina; Dwyer 1958, Sprada Tavares andda Silva 1992), mainland tropical Africa (equatorial forest belt; Léonard 1951), Madagascar and the Comoros Islands (Du Puy et al. 2002) and the Indo-Pacific (extending from the Western Ghats eastward to Fiji; Knaap-van Meeuwen 1970, Smith 1985. Perhaps unsurprisingly, the genus has long troubled taxonomists and systematists who have struggled with its diversity in several attempts to revise and classify regional groupings of the species (Léonard 1951, Dwyer 1958, Knaap-van Meeuwen 1970. Further, its relationship to the much smaller Pacific genus Maniltoa Scheff. has remained equivocal. Several important phylogenetic studies of the Caesalpinioideae sensu lato suggested that Cynometra was not monophyletic (Bruneau et al. , 2008. More recent studies, with greater taxonomic sampling, have provided additional evidence that the genus is polyphyletic and, furthermore, that one of the lineages is paraphyletic with respect to Maniltoa (Mackinder et al. 2013, Radosavljevic et al. 2017, de la Estrella et al. 2018. In order for the classification of Cynometra to reflect a monophyletic taxon, a new genus circumscription must be provided and Maniltoa should be subsumed into Cynometra. In addition, during the course of herbarium study and fieldwork, several undescribed species were revealed. Those species are here described. This paper is divided into three parts: I. A new circumscription of the genus Cynometra; II. The transfer of all species of Maniltoa into Cynometra; and III. Description of four new species of Cynometra.

Materials and methods
The following herbaria allowed the author to study their collections in person or provided loans: A, ATH, BM, BRI, CANB, F, IAN, INPA, K, MG, MO, NY, P, RB, SING and US. Morphological characters were assessed and measured from herbarium material. Flowers and immature fruit were rehydrated in warm Pohl's solution (Pohl 1965) prior to dissection and measurement. Other characters were measured directly from the herbarium specimens. For the new species from Ecuador, flower colour, habit and habitat information were taken from label data and field observations by Gwilym Lewis and Bente Klitgaard (pers. comm.); for the other three species, these features were described from label data. Reproductive status is indicated with the following abbreviations (alone or in combination): fl, flowering; fr, fruiting; im, immature; st, sterile. Illustrations were made at the National Museum of Natural History, Washington DC, USA by Mattias S. Lanas (Cynometra steyermarkii), Katherine Rudebusch (C. tumbesiana) and Alice Tangerini (C. cerebriformis and a.  Radosavljevic et al. (2017) with branches coloured by species distributions. Star indicates type species of Cynometra C indehiscent fruits typical of Cynometra as circumscribed here; D dehiscent fruits typical of those species formerly included in Cynometra, but excluded by the new circumscription.
C. dwyerii) from herbarium specimens. Two of the new species are described based on a single collection: in both cases, the taxa are morphologically and geographically distinct and, after extensive herbarium study, the author is confident that the available material, while limited, is sufficient to distinguish two new entities at the species level.

Part I: A new circumscription of the genus
Linnaeus established the genus Cynometra and included two species from southeast Asia in the first edition of Species Plantarum, Cynometra cauliflora L. and C. ramiflora L. (Linnaeus, 1753). This original concept of Cynometra was of a genus distinguished by compound leaves, four reflexed sepals, five equal petals and by single-seeded, indehiscent pods with thick and tuberculate valves (Linnaeus' description does not explicitly mention 'indehiscent pods', but it is a character present in both of his original species). For nearly the next century, this circumscription was generally followed by taxonomists around the world. The limits of the genus were first expanded by the publication of two new species from the Indo-Pacific region: Cynometra polyandra Roxb., described in 1820 from Indian material and Cynometra grandiflora A.Gray, described in 1854 from Fijian specimens. These species differed most notably from the traditional understanding of Cynometra by having more numerous stamens (35-40 in C. grandiflora and approximately 50 in C. polyandra; Roxburgh 1820, Gray 1854). Bentham (1865a) later described the African species C. floribunda Benth. and C. laxiflora Benth., species with large, petaloid bracteoles and only two showy petals which he placed in a new subgenus, Hymenostegia (the fruit of the two species also differs from Cynometra, but these were unknown at the time). Scheffer (1876) established the segregate genus Maniltoa after viewing collections from New Guinea; believing the material to be conspecific, he included a single species, Maniltoa grandiflora (A.Gray) Scheff. (based on Cynometra grandiflora A.Gray). While the most apparent difference between the new genus Maniltoa and typical Cynometra was the number of stamens, Scheffer (1876) argued that the genus was distinct based on the size of the flowers, the number of seeds, caducous filiform stipules, the presence of a tendril terminating the leaf rachis and bracts subtending each pedicel in the inflorescence. He does not include or reference C. polyandra, another polyandrous species. Maniltoa was not universally accepted at first and Taubert's (1894) treatment of the Leguminosae for Engler's Die Natürlichen Pflanzenfamilien treated Maniltoa as a section of Cynometra (along with the sections Eucynometra, Hymenostegia and Pseudocynometra). A few years later, however, Harms (1897) broke apart this broad concept of Cynometra. He elevated Hymenostegia to the rank of genus and restored Maniltoa by combining it with section Pseudocynometra (consisting of a single species, C. polyandra) on the basis of greater stamen number and the presence of prominent buds enclosed in imbricate scarious bract scales (Harms 1897(Harms , 1902(Harms , 1919. This understanding of Cynometra and Maniltoa in the Indo-Pacific was further refined by Knaap-Van Meeuwen (1970) who sought to better define the limits between the two genera -she again placed greatest significance on the stamen number, placing those with 8-13 stamens in Cynometra and those with 15-80 in Maniltoa. While many authors have followed this delimitation (Verdcourt 1979;Hou et al 1996), some have continued to question the distinctionof the two genera (Soerianegara 1993;Pan et al. 2010).
In Africa, misunderstanding of the delimitation of Cynometra persisted, even after Harm's (1897) treatment and the turn of the century marked the beginning of a period of rapid taxonomic expansion in both the number of species and, consequently, the characters describing the generic limits of Cynometra. As explained (in great detail) by Léonard (1951), the first few species described in Africa generally matched the Linnaean concept (e.g. Cynometra vogelii Hook.f. in 1849 from Nigeria; C. mannii Oliv. in 1871 from Cameroon). However, with the description of Cynometra sessiliflora Harms, Harms (1899) expanded the understanding of Cynometra to include species with five erect sepals (instead of four reflexed sepals) and, most importantly for the continued confusion regarding the genus, dehiscent pods with thin, flat valves. Not long after, the publication of Cynometra alexandri C.H. Wright, C. gilletii De Wild., C. laurentii De Wild., C. lujae De Wild., C. oddonii De Wild. and C. pedicellata De Wild. added more species with five sepals and dehiscent fruits and further expanded the genus to include species with an intrastaminal disc and numerous ovules (Wright 1902;De Wildeman 1904, 1905, 1906. Harms (1907) also continued to expand the generic conception of Cynometra in Africa to include species with alternate leaflets, distichous flowers and a stipe fused with the receptacle wall when he described C. leptantha Harms, C. multijuga Harms and C. pierreana Harms. Cynometra brachyura Harms, expanded the genus to again include species with only two petals (Harms 1913). As a result, species with all manner of combinations of these characters were placed in Cynometra (see Baker 1930 andPellegrin 1949). While many taxonomists, working in Africa, noted the heterogeneity of the genus, it was not until Léonard undertook a thorough and careful revision of the African species that a coherent circumscription of Cynometra began to emerge (Baker 1930;Lebrun 1933;Aubréville 1936;Léonard 1951Léonard , 1957. Léonard's work (1951Léonard's work ( , 1957) was a major advancement in revising the boundaries of Cynometra and its allies in Africa, which had become so broad as to make genera nearly indistinguishable. Léonard (1951) removed from Cynometra any taxa with the stipe of the ovary adnate to the hypanthium or eccentrically inserted (species with a free, central stipe but alternate leaflets with translucent dots were moved to Gilletiodendron Vermoesen). To accommodate his revised concept of the genus, Léonard (1951Léonard ( , 1957) described a new genus, Lebruniodendron J.Léonard, resurrected the genera Gilletiodendron Vermoesen and Zenkerella Taub. and transferred numerous species out of Cynometra and into the genera Hymenostegia (Benth.) Harms, Plagiosiphon Harms, Schotia Jacq. and Scorodophloeus Harms. He placed the remaining Cynometra into three informal groups; they share a centrally inserted stipe, opposite leaflets, five equal petals and imbricate sepals, but differ notably in characters such as inflorescence structure, presence or absence of a staminal disc, presence or absence of foliar glands and fruit shape and dehiscence. Léonard (1951) acknowledged that the genus remained heterogenous -the species of 'group one' resembled 'typical Cynometra', but the others were possibly misplaced. Later, he further assessed the genera of African Amherstieae and Cynometreae, revising the tribal and generic limits using characters from germination mode, seedling architecture and wood structure, amongst others. (Léonard 1957). A thorough examination of the genus Cynometra, unfortunately, was not included in that otherwise exhaustive study. Léonard and others recognised that, as contemporarily circumscribed, the genus was likely polyphyletic and would require further revisionary studies at a global scale (Aubréville 1968(Aubréville , 1970Breteler 1996;Léonard 1996).
An early morphological cladistic analysis by Temu (1990) provided the first phylogenetic evidence for the non-monophyly of African Cynometra. The results showed two distinct clades of Cynometra in tropical Africa, one of which included Maniltoa. Later morphological and molecular phylogenetic analyses, albeit with limited sampling of Cynometra and Maniltoa, also found no evidence to support monophyly of Cynometra and suggested that it may be paraphyletic with respect to Maniltoa (Bruneau et al. , 2008. More recent studies which included a greater taxon sampling from Cynometra and Maniltoa have found support for two clades of Cynometra, one consisting of exclusively East African species and the other containing species from the rest of the genus plus Maniltoa (Mackinder et al. 2013;Radosavljevic et al. 2017;de la Estrella et al. 2018). Radosavljevic et al. (2017), which featured 47 accessions from 36 species of Cynometra and Maniltoa, recovered two strongly supported clades of Cynometra sensu lato, which are each diagnosed by several morphological characters ( Figure 1). One clade (Cynometra clade A of Radosavljevic et al. 2017) is comprised of exclusively east African species and is grouped as sister to a clade composed of Dicymbe Spruce ex Benth. and Polystemonanthus Harms with strong support. The other clade, which includes the type species Cynometra cauliflora L., contains Cynometra species from the Neotropics, the remaining Afrotropical species and the Indo-Pacific species (Cynometra clade B of Radosavljevic et al. 2017). Maniltoa is nested within clade B and is also recovered as non-monophyletic. This clade of Cynometra and Maniltoa is placed in a clade with Zenkerella Taub., Normandiodendron J. Léonard and the Scorodophloeus clade, again with strong support. The species of Cynometra clade A are characterised by the absence of foliar extra-floral nectaries, paniculate inflorescences (with two exceptions), articulated pedicels, a well-developed intrastaminal disc, 2-4 ovules and dehiscent fruit. Cynometra clade B is characterised by the presence of foliar extra-floral nectaries (in most species), racemose inflorescences, simple pedicels, the absence of an intrastaminal disc, 1-2 ovules and indehiscent fruit. The species of Cynometra clade A correspond to Léonard's (1951) groups 2 and 3 while Cynometra clade B, which contains the generitype Cynometra cauliflora L., corresponds to group 1. Léonard (1951Léonard ( , 1996 predicted as much; he referred to group 1 as 'typical Cynometra' and believed that Cynometra alexandri and C. hankei Harms were likely misplaced. De la Estrella et al. (2018), who sampled broadly across Detarioideae, placed Cynometra and Maniltoa in a re-circumscribed Amherstieae.
To reconcile the discrepancy between the current classification of Cynometra and Maniltoa and our understanding of their evolutionary relationships, I present here an emended generic circumscription of Cynometra that excludes the species of Cynometra clade A and includes the species formerly placed in Maniltoa. A treatment that proposes a new genus for the species of Cynometra clade A is in preparation. Trees or shrubs (infrequent), evergreen (rarely deciduous), buttressed or not, growth flush-wise, new growth flaccid and whitish or reddish at emergence, becoming green with maturity, vegetative buds covered in a series of imbricate scales. Stipules lateral, free, linear or filamentous, early caducous, scars typically not visible on mature growth. Leaves pulvinate, petiolate, rachis (if present) terete or caniculate, often terminating in a filiform outgrowth, axes glabrous or pubescent, paripinnate with 1-16 pairs of opposite leaflets (rarely unifoliolate); leaflets petiolulate but sometimes appearing sessile because of decurrent lamina, blade lanceolate, ovate, elliptic, oblong, obovate, oblanceolate or trapeziform, symmetrical to strongly asymmetrical, glabrous or sparsely pubescent abaxially, glabrous adaxially (rarely with sparse pubescence along midvein), margins entire, apex broadly obtuse to acuminate, retuse or emarginate, mucronate, base oblique, distal margin decurrent to petiolule; laminar nectaries usually present, abaxial, submarginal, embedded in laminar surface, shallow, without noticeably raised edges. Inflorescences axillary or ramiflorous (rarely cauliflorous), racemose, 1-2(-3), per axil, buds enclosed by imbricate bracts, appearing conical, ovoid or cigar-shaped in silhouette; pedicels simple (not articulated), filamentous in anthesis, accrescent and lignified in fruit; bracts enclosing inflorescence during development prior to emergence, imbricate, distichous, deciduous or persistent; bracteoles not enclosing buds, paired, inserted along proximal half of pedicel, opposite or subopposite, caducous. Flowers bisexual, actinomorphic; hypanthium present (sometimes indistinct), either turbinate-campanulate and short or tubular and partially to completely enveloping ovary; disc absent; sepals (3-)4(-5), reflexed at anthesis, unequal, deltoid, ovate, elliptic or oblong; petals 5, crumpled or smooth, incurved, erect or horizontal, equal, linear, oblong, or oblanceolate, with rounded or acute apices; stamens (8-)10-80, filaments free or briefly connate basally; anthers dorsifixed, versatile, longitudinally dehiscent; ovary inserted +/-centrally, occasionally eccentric, free, short-stipitate or subsessile, pubescence varied but rarely glabrous, light green or pinkish, often turning red post anthesis, ovules 1 (2); style eccentric, glabrous; stigma capitate. Legumes indehiscent, splitting along suture after germination due to action of emerging shoot, orbicular to oblong, often laterally compressed, smooth to deeply fissured, often apiculate (in some taxa only when immature), valves 0.5-4.0 mm thick, cork-like in several species, remaining attached to cotyledons during and after germination. Seeds 1-2, enclosed in fruit until germination. Seedling germination epigeal. The emended description above reflects the merger of Maniltoa and Cynometra s.s. and the exclusion of the east African taxa with dehiscent pods, paniculate inflorescences and articulated pedicels (Tables 1, 2). While many taxa in the former Maniltoa are no doubt distinctive and, in many cases, striking when compared to the more unassuming Cynometra, a close examination of their morphological characteristics reveals a strong similarity. The number of stamens has been perhaps the most distinctive and most often cited character separating the two genera. Cynometra has long been known to have only 10 stamens, while Maniltoa has been traditionally treated as having anywhere from 15-80 stamens. Indeed, this proliferation in stamen number is rare in the Detarioideae. However, a closer look at the Indo-Pacific Cynometra reveals several examples of species which regularly have as few as eight stamens (Cynometra glomerulata Gagnep.) or as many as 12 (Cynometra katikii Verdc.) and some with variation across the range within a species (8-10 stamens in Cynometra cauliflora L. and 10-15 stamens in Cynometra ramiflora L.; Knaapvan Meeuwen 1970, Verdcourt 1979. So, while the majority of species, placed within Cynometra, do have 10 stamens, there is lability amongst this trait in the taxa of the region. Several other characteristics of Maniltoa that supposedly distinguish it from Cynometra are also incorrect. The first is the presence of conspicuous 'bract covered' vegetative and reproductive buds (Scheffer 1876, Harms 1902, 1919. In some species of Maniltoa, these buds can be many centimetres long and several centimetres in diameter. The scales themselves are often tan or brown, but can be whitish or pinkish and are often covered with parallel striations running longitudinally along the surface. These types of buds are also present in Cynometra, although they are much smaller, in some cases only 4 or 5 mm long. Likewise, the scales are similar in shape, vestiture and surface texture. Another general characteristic used to separate the two genera has been the sturdiness of the inflorescence rachis in Maniltoa (Knaap-van Meeuwen 1970). However, this is likely related to the overall general difference in size of the flowers and inflorescence between the two genera: Maniltoa tend to have larger buds, bracts, inflorescences and individual flowers. The larger inflorescences of the Maniltoa species may be an adaptation to mammal pollination as there are reports of marsupials and bats visiting the flowers of Maniltoa species in Australia and the Pacific Islands (Marshall 1985, Endress 1994). Scheffer (1876) cited the presence of a tendril terminating the leaf rachis as a distinguishing character of Maniltoa, but Dwyer (1958) notes that this trait is present in several Cynometra. Finally, differences in wood anatomy given by Knaap-van Meeuwen (1970) have been found to be insufficiently distinct from one another and overlapping in range (Soerianegara 1993;Pan et al. 2010).
The most clearly distinguishing feature of the newly emended Cynometra with respect to the taxa here excluded (Cynometra clade A) is the indehiscent pod. Indeed, if one examines the taxa misplaced in Cynometra over the years, they nearly all share the characteristic of dehiscent pods. Unfortunately, Linnaeus makes no mention of the  (1897) all described the pods as 'two-valved' without further elaboration. Léonard (1951) and Aubréville (1968Aubréville ( , 1970 cite the pods of 'typical Cynometra' as dehiscent or indehiscent, likely owing to the fruit of Cynometra mannii, the valves of which are not dehiscent but, upon pressing and drying, often split and rupture in several places along axes roughly perpendicular to the sutures. A similar occurrence is observed in the Neotropical species Cynometra bauhiniifolia Benth. When Harms (1899) described C. sessiliflora, he set a precedent regarding the pod characters that contributed to the imprecise delimitation of Cynometra and subsequently led to the taxonomic confusion surrounding Cynometra in Africa. With this new circumscription, the two genera are united and share radially symmetric flowers, bract covered buds (with bracts persisting on the inflorescence), flowers with early caducous bracteoles not enveloping the flowers and indehiscent fruits (which may play a role in dispersal as many taxa are associated with alluvial habitats or appear as drift 'fruit' in coastal environments; Ridley 1930, Léonard 1951, Clarke et al. 2001, Tomlinson 2016. The new classification changes the overall distribution of the taxa. Prior to this, Cynometra was a genus with its diversity distributed somewhat evenly across the American, African and Asian tropics. However, with the exclusion of many of the African species and the inclusion of the former Maniltoa, Cynometra is now a genus whose primary centre of diversity is the Indo-Pacific region, with a secondary centre in the Neotropics and the majority of Afrotropic taxa restricted to Madagascar. Notes. The arrangement of leaflets for which this species is named (i.e. the basalmost pair of leaflets inserted just above petiole and separated from terminal pair of leaflets by a relatively long rachis) is unusual amongst the species that were formerly included in Maniltoa; however some species of Cynometra have a similar arrangement (e.g. Cynometra sakalava Du Puy & R.Rabev from Madagascar). Notes. The protologue lists the collection year as 1928, however 1926 appears to be the correct date. While the label affixed to the holotype does list the collection date as "14 May 1928", this appears to be a transcription error. The handwritten slip attached to the holotype, presumably filled out by the collector Brass, gives the date as "14/5/26". Additionally, the printed label is titled "Arnold Arboretum Expedition, 1925Expedition, -1926. The label on the isotype also has the collection date as 14 May 1926. ( Notes. According to Knaap-van Meeeuwen (1970, p. 42), the type material for Maniltoa browneoides came from a cultivated specimen grown at Bogor Botanical Gar-dens from seeds or seedlings collected by Forbes on New Guinea, however no citation is given as the source for this information. The holotype for M. gemmipara is listed as being at Leiden by Knaap-van Meeuwen, however there are no specimens there annotated as such. There are two specimens, however, from Java collected in 1908 that seem to match the details in the protologue. These are filed as M. gemmipara and have the registration numbers L.3886447 and L.3886448.  Note. When Verdcourt (1977) published the above name at a new rank, he did not name a neotype. He did cite several representative specimens and specifically men- tioned Coode et al NGF 29604 as a specimen he was certain belonged to this species. Given that it is well represented in herbaria and was collected near the original type locality, this specimen has been chosen as the neotype. Notes. This species is named after the nation of Fiji, where it is endemic, but somewhat common and widespread, occurring on at least six islands (Smith 1985).

Cynometra cerebriformis
Distribution and ecology. Currently, this species appears restricted to the Trombetas River Basin of Para State, Brazil. However, it is quite possible that the range is more extensive given that the species occurs in seasonally flooded forests and the genus is known for water dispersed fruits (Ridley 1930, Clarke et al. 2001, Tomlinson 2016. Little else is known about this taxon. Phenology. Flowering specimens have been collected in May. Fruiting specimens have been collected in June and August.  Notes. This taxon has been collected in the areas around Santarem and Oriximiná in Para, Brazil. Many of the specimens have been annotated by Adalea Sprada Tavares as Cynometra duckei ssp. trombetensis, but I can find no record of publication and several of the specimens thus annotated are assigned to different taxa in her unpublished thesis (Sprada Tavares 1987). The available material of C. duckei is limited and only a few fruiting specimens exist. While a case can be made that C. duckei shares certain vegetative traits with C. cerebriformis (prominent lenticels, the smoothness of the adaxial surface of the leaflets), it also shares characteristics with several other taxa, including C. spruceana var. spruceana (long pedicels, leaflet shape) and C. marginata var. laevis (smooth leaflet surface, nearly sessile leaflets). In light of the characters separating C. cerebriformis from other taxa (see below), the author has opted to describe this taxon at the species level.
Cynometra cerebriformis differs from C. duckei primarily in the shape of the leaflets. Cynometra cerebriformis differs from C. spruceana var. spruceana in several ways. The leaflets of C. cerebriformis are generally smaller than those of C. spruceana var. spruceana and the surface is nearly smooth, while the secondary veins are quite obvious in C. spruceana. Cynometra cerebriformis also lacks the basal laminar gland present in C. spruceana and many other Cynometra taxa. Finally, the fruit of C. spruceana is approximately 1.5×-2.5× larger than the fruit of C. cerebriformis and the valves are smooth to slightly rugulose, lacking the strongly rugose surface of C. cerebriformis. Cynometra cerebriformis differs from C. marginata var. laevis by having an acute leaflet base, short acumen and oblate rugose fruit; C. marginata var. laevis has an obtuse leaflet base, long acumen and a globose fruit with a prominent raised suture ridge. Description. Tree to approximately 20 m tall; bark not seen; branchlets lenticelate, pubescent when young, becoming glabrous with age. Stipules not seen. Leaves bifoliolate, axes pubescent, transversely corrugated; petioles 4.5-5.5 mm long; petiolules 1.0-1.5 mm long, inconspicuous, leaflets appearing sessile; leaflets narrowly obovate to obovate, occasionally sub-trapeziform, strongly asymmetric, primary vein  eccentric, proximal side 2.8-3.7 times wider than distal, 3.1-3.9 cm long, 1.4-1.9 cm wide, thin, abaxial surface sparsely pubescent, more so on midvein and major secondaries, adaxial surface with pubescence restricted primarily to midvein, occasional hairs scattered on lamina, primary venation pinnate, secondary venation brochidodromouseucamptodromous, 2(-3) basal acrodromous veins, decurrent to primary, prominent abaxially, only slightly less so adaxially, tertiary venation visible on both surfaces at 10× magnification, margins entire, apex acute, weakly acuminate (to 2.0 mm), retuse, mucronate, base oblique, acute, distal side narrowly cuneate, proximal side slightly concave to cuneate, decurrent to petiolule, laminar glands present, 3-6 per leaflet, arranged in a row approximately halfway between margin and midvein, restricted to distal portion of lamina, typically adjacent to tertiary veins, crateriform, less than 1.0 mm in diameter. Inflorescences not seen, position inferred as axillary from remnant of peduncle. Flowers not seen. Legume indehiscent, roughly globose, to 4.7 cm in diameter, surface of valves rugulose, wall of pericarp up to 3.0 mm thick, deep brown colour at maturity. Seeds 1 per pod, filling entire cavity, dark brown.

Cynometra dwyerii
Distribution and ecology. Known only from the type locality in the Darien Gap region of Panama; the area where the type was collected is primarily lowland moist tropical forest.
Phenology. The type was collected with mature fruit in July. Etymology. The specific epithet honours the contributions of Dr. John Dwyer, who published the first monograph of the Neotropical species of Cynometra.
Note. This species is known only from the type collection, however the combination of vegetative characters and fruit morphology make it clearly distinct from other Cynometra species. Superficially, this species resembles C. bauhiniifolia, given its small leaflets with prominent secondary venation. However, the pubescence and arrangement of laminar glands clearly distinguish it from other Neotropical species of Cynometra, which usually have just a single, basal laminar gland. In fact, the combination of bifoliolate leaves with small leaflets and several submarginal laminar glands is unique across the entire genus. When these characters are combined with the large, globose fruit, it is clear that this is a distinct species.

Cynometra steyermarkii
Distribution and ecology. Known from a single collection made in Miranda State, Venezuela. The species was collected growing along a small stream in the foothills of the Cordillera de la Costa at approximately 75 m above sea level. While much of this region is covered in xeric shrublands and thorn forests, this site corresponds to a low gallery forest growing along a tributary of the Río Aricagua.
Phenology. The type material was collected at the end of March with late flowers and nearly mature fruit.
Etymology. The specific epithet honours Dr. Julian A. Steyermark, the collector of the type material and a prolific collector of neotropical plants. Over the course of his   Notes. Cynometra steyermarkii is the only confirmed species of Cynometra known from the forests of the Cordillera de la Costa in northern Venezuela and one of two species of Venezuelan Cynometra growing outside of the Guayana region. Henri Pittier (1926) published an account of a Cynometra growing in the hills outside of Caracas, Cynometra sphaerocarpa Pittier, for his Manual de las plantas usuales de Venezuela. While the description is valid, according to the rules of nomenclature in effect at the time, he failed to designate a type or list any collections that may correspond to this taxon. In his 1958 revision, Dwyer was unable to locate any material that could be assigned to this taxon and pointed out the limited utility of Pittier's description -'Las hojas inequilaterales, lanceadas, largamente atenuadas y glabras… Los frutos son subglobosos, de 3.5-4 cm. de diametro y contienen una sola semilla [Leaflets asymmetric, lanceolate, largely attenuate and glabrous. The fruits are sub-globose, 3.5-4 cm in diameter, and contain a single seed].'-which could very well describe any number of Cynometra species, including C. steyermarkii. A full set of Pittier's duplicates from this publication should be deposited in the United States National Herbarium (US), but after exhaustive searches in the US collections and a study of both Pittier's archives and the museum registrar's records, I was unable to find any material that could be attributable to C. sphaerocarpa. Curators at Herbario Nacional de Venezuela graciously searched through material housed there, but were also unsuccessful. While it is possible that C. steyermarkii and C. sphaerocarpa are the same taxon, without Pittier's original material, it is impossible to know for sure. Rather than leave this unresolved, it is more useful to describe a new species with good type material.

Cynometra tumbesiana
Distribution and ecology. Cynometra tumbesiana occurs in the seasonally dry tropical forests of western and southern Ecuador and north-western Peru (a single collection) at elevations between 100-800 m (Figure 3). These habitats are severely threatened regionally and globally due to extensive human modification of the landscape and, as a result, this species now exists primarily as isolated fragments. While C. tumbesiana is locally abundant at a few sites, it is currently known from less than 10 localities, several of which are within 5 km of each other.
Phenology. Flowering specimens have been collected in May and December to January; fruiting specimens have been collected in August and January to February. Cynometra tumbesiana is one of the few woody taxa in the dry forests that retain their leaves during the dry season.
Etymology. The specific epithet refers to the Tumbes region, where the type specimen was collected and where many of the known localities occur. Cynometra tumbesiana is morphologically and ecologically similar to C. oaxacana Brandegee from western and southern Mexico. The two can be distinguished by the narrower and more acuminate leaflets and larger fruits in C. tumbesiana. The inflorescences of C. oaxacana are also slightly more robust and have a more obvious pubescence. Both species are found in dry habitats (uncommon amongst the neotropical species of Cynometra), though C. tumbesiana is found in much drier sites.