Commelina Plum.

Genus Commelina Plum. ex L. The gem-fruited Commelina Athyrocarpus Schltdl. (Schlechtendal 1855: 454); Hasskarl (1866: 212); Bentham in Bentham & Hooker (1883: 847). – Type species: Commelina pallida Humb. & Bonpl. ex Willd.; designated here. Commelinopsis Pichon (Pichon 1946: 227), pro parte. – Type species: Commelina persicariifolia Redouté. Phaeosphaerion Hassk. (Hasskarl 1866: 212). – Type species: Commelina leiocarpa Benth. Characterisation and circumscription This informal and non-monophyletic group is treated as such due to the historical taxonomic and nomenclatural confusion involving its species. The only character uniting these eight species is their showy and gem-like fruits, used in the past to delimit genera previously segregated from Commelina (i.e., Athyrocarpus, Commelinopsis, and Phaeosphaerion) and tribe Pollieae C.B.Clarke (Clarke 1881). Despite the morphological dissimilarity between these species, they continue to be confused and lumped together by field and generalist botanists, as well as by ecologists, parataxonomists, and amateurs. Thus, treating them in the same study not only makes sense from a nomenclatural and taxonomic perspective but also centralises the most up-to-date information available for this group. The fruits in the gem-fruited Commelina can range from dehiscent to partially dehiscent to indehiscent, crustaceous or not, and be shiny pearly-white to silvery, opaque off-white, opaque reddish-brown to dark maroon, or glaucous atro-vinaceous to bluish-black to black. The seeds can be monomorphic or dimorphic, free or variously fused to the fruit wall or septa, and their ornamentation can range from smooth to variously ornate. However, as hypothesised by Faden & Hunt (1987), the gem-like fruits seem to have evolved independently several times within Commelina since all gem-fruited species are morphologically and phylogenetically related to other species that lack such fruits (Pellegrini et al. in prep.). Furthermore, gem-fruited species of Commelina are recovered in at least three distantly related lineages in the genus (Pellegrini et al. in prep.). Accepted species We currently accept eight species of gem-fruited Commelina, plus six morphologically similar or closely related species, totalling 14 species. These species are organised into three morphologically cohesive and molecularly supported species groups (Pellegrini et al. in prep.), namely: 1) Commelina benghalensis group [i.e., C. obliqua Vahl, C. pseudomonosperma (Kuntze) L.M.Campb., and C. rufipes Seub.]; 2) C. robusta group [i.e., C. bambusifolia Matuda, C. huntii M.Pell., C. robusta Kunth, C. scabrata Seub., C. sugariae M.Pell. sp. nov., and C. vestita Seub.]; and 3) C. tuberosa group [i.e., C. almandina M.Pell. & Cornejo sp. nov., C. efoveolata (C.B.Clarke) L.M.Campb., C. leiocarpa Benth., C. pallida Humb. & Bonpl. ex Willd., and C. texcocana Matuda]. Distribution This non-monophyletic group is currently restricted to the Neotropics, although at least one undescribed species is from the Palaeotropics (i.e., Madagascar; Faden & Hunt 1987). In the Neotropics, the group occurs from Mexico, reaching the Antilles and extending all the way down to northern Argentina, Bolivia, and Brazil. Brazil and Mexico are the most species-rich countries for members of this group, with 8 and 7 species, respectively. Ecology Most species in this group grow in the understory of dry or rainforests, with a few of them also occurring or being restricted to open environments. Fruit dispersal Fruit and seed dispersal in the gem-fruited species has never been studied, and this reflects the complete absence of dispersal studies for Commelinaceae (Pellegrini & Faden 2017). However, it is hypothesised that the species with indehiscent fruits are zoochoric and potentially dispersed by birds (Faden & Hunt 1987; Pellegrini & Faden 2017). However, the species with dehiscent fruits (e.g., the newly described C. almandina M.Pell. & Cornejo sp. nov.) seem less likely to be dispersed by birds, despite being dark-coloured, clearly exposed by the plant’s vining habit, and being exerted from the spathe during maturity (Pellegrini, pers. obs.). Species with white fruits (dehiscent or not) are small-sized, prostrate to ascending, sometimes scrambling, small understory herbs, making their fruits somewhat difficult for most open environment birds to access. This makes it more likely for the dispersers to be some type of understory bird(s) native to the Neotropical rainforests (Pellegrini, pers. obs.). Nomenclatural remarks Athyrocarpus was first described by Schlechtendal (1855: 454), who provides a rather detailed diagnosis for his new genus, characterising it by its sub-solitary inflorescences, with the superior cincinnus aborted or completely lacking and the lateral one 2–4-flowered, flowers similar to Commelina in possessing 5 stamens (2 of which are ended by antherodes, i.e., staminodial), indehiscent green fruits becoming lead-blue at maturity, and dimorphic seeds. Schlechtendal (1855) goes into further detail, comparing his new genus to Aclisia E.Mey. ex C.Presl, Pollia Thunb., and Lamprocarpus Blume (all currently treated as Pollia s. lat.) due to their similar fruit morphology. Therefore, Schlechtendal’s (1855: 454) is considered the original work where this name was first validly published. Thus, Hasskarl (1866: 212) and Bentham & Hooker (1883: 847) actually represent citations of Schlechtendal’s genus and not attempts to validate the name as considered by many databases. Finally, Schlechtendal (1855: 454) included two species in his new genus, C. pallida and C. rubens Redouté; however, without formally transferring any of them to Athyrocarpus. Based on Schlechtendal’s diagnosis, plus the fact that C. rubens is considered to be distantly related to C. pallida (this study; see Remarks under C. rubens and Discussion), we designate C. pallida as the type species for Athyrocarpus. Interestingly, neither C. pallida nor C. rubens present the gem-like blue fruits deemed to be diagnostic for the genus. When describing Commelinopsis, Pichon (1946) unambiguously designated Commelina persicariifolia Redouté as the type species for his new genus by including it as its sole species. Nonetheless, the species does not possess the crustaceous, indehiscent and pearly-white to silvery capsules and seeds fused to the fruit wall and septae, which were listed by the author as diagnostic for the genus (Pichon 1946: 227). This is easily explained by Pichon’s incorrect application of C. persicariifolia, which for Pichon was conspecific with C. rufipes. Therefore, despite the widespread use of this generic name for the C. rufipes complex, it is actually typified by an unrelated species belonging to the C. tuberosa group. Phaeosphaerion was described by Hasskarl (1866: 212) based on his identification key for the genera of Commelinaceae. Despite being brief, the step leading to Phaeosphaerion characterises the genus as having flowers obscured by the spathe, simple stigma, and indehiscent, shiny and lead-black capsules. Despite not formally transferring it to his new genus, Hasskarl (1866: 212) does cite C. leiocarpa Benth. Hence, this species must be treated as the type species for the genus Phaeosphaerion. Key to the gem-fruited Commelina Plum. ex L. and allies 1. Leaf-sheaths with hyaline to white hairs along the margin; inflorescences predominantly axillary and leaf-opposed, long pedunculate (peduncle the same length or longer than ½ length of the spathe), spathe base free, internally inconspicuously mucilaginous; sepals opaque, green, lower sepals sessile, free, medial petal shortly-clawed, style base tapering into the ovary; fruits thick-walled ................. ............................................................................................................ 2 (Commelina tuberosa group) – Leaf-sheaths with light brown to brown or rusty to rusty-brown or red to dark red to atro-vinaceous hairs along the margin; inflorescences terminal or apparently so, short-pedunculate to sessile (peduncle shorter than ½ length of the spathe), spathe base connate at least at base, internally conspicuously mucilaginous; sepals hyaline, colourless to white, lower sepals shortly-clawed, connate up to half their length, medial petal sessile, style base not tapering into the ovary; fruits thin-walled ............ 6 2. Peduncle pendulous, spathe slightly falcate, base cordate; pedicels puberulous with minute hook-hairs only, style twice as long as the stamens, persistent in fruit, stigma tan-coloured; fruits indehiscent, glaucous, not constricted between the seeds when mature, atro-vinaceous to bluish-black to black when mature, apex round to apiculate; seeds monomorphic, triangular to widely triangular to triangular-ellipsoid ............................................................................................................................................. 3 – Peduncle straight, spathe straight, base subcordate to round; pedicels pilose with acicular hairs or with a mixture of acicular and hook-hairs, style slightly longer than the stamens, deciduous in fruit, stigma white; fruits dehiscent, 3-valved, opaque, constricted between the seeds when mature, tan-coloured or reddish-brown to dark maroon when mature, apex rostrate; seeds dimorphic, ellipsoid ........................................................................................................................................................... 4 3. Stems branched from the base or almost so, sparsely to densely hispid with acicular hairs; leaf-blades adaxially sparsely to densely hispid with acicular hairs, abaxially sparsely to densely hispid with acicular hairs, hairs generally congested along the midvein, margin ciliate with a mixture of prickle- and acicular hairs; spathe externally sparsely to densely hispid with acicular hairs; dorsal sepal pilose along the midvein, lower sepals widely elliptic, petals whit