Psychotria involucrata of authors, not Sw.
Good flowering material of this species is rare, most of the specimens having been collected in the fruiting stage only. In early anthesis the inflorescence in most specimens appears compact and monocephalous (Gleason 931 from Tumatumari) with the outer bracts proximate, but in post-anthesis and in fruit there is a gradually loosening transitional series of stages involving the breaking up of the axis into 4-6 short axes, at first scarcely elongated to 2 mm (de la Cruz 3236, Sandwith 1172), eventually elongating to 3 mm (Sandwith 1172) or to 6-10 mm (de la Cruz 3236) in the fruiting stage. The bracts thus lose their original character passing into fruiting stages.
This taxon illustrates the difficulty in distinguishing the concepts of Cephaelis from Psychotria. The Cephaelis-like character of the involucral bracts is maintained in the early stages of anthesis, but this character becomes less and less conspicuous as the fruiting stage is reached. Müller-Argoviensis placed this species in sect 10 Cephaelis of Psychotria, while Bremekamp regarded it as a true Psychotria, and N. E. Brown at Kew interpreted it as Cephaelis (Gleason 931 from British Guiana.).
Psychotria barbiflora A.DC. is the first available name for the plant passing as Psychotria involucrata of authors, not Swartz. Swartz’s P. involucrata (Prodr. Nov. Gen. 45. 1788) was a superfluous name for Nonatelia officinalis Aubl. (Hist. PI. Guian. 182 t. 70, fig 1. 1775), and was taken up by Willdenow (Sp. PI. 1: 965-966. 1798) who referred to the Jamaican reference of Shakespear’s specimen in Swartz’s Prodr. Fl. Ind. Occ. 1: 413. 1797, and to the Guiana specimen of Aublet’s Nonatelia officinalis. Steudel, likewise, in his Nomenclator Botanicus p. 665. 1821, synonymized Nonatelia officinalis Aubl. and Psychotria officinalis (Aubl.) Raeusch. as pertaining to Psychotria involucrata Sw. As such, Psychotria involucrata Sw. has been identified with P. officinalis (Aubl.) Raeusch., but Sandwith (Kew Bull. 1931: 473. 1931) identified P. involucrata Willd., not Sw., with P. officinalis.
Urban (Symb. 7: 473. 1913) attempted to retain P. involucrata Sw. (excluding syn. Aubl.) and P. hoffmannseggiana together as the same species, including the Cuban P. tribracteata C. Wright, citing the Shakespear collection from Jamaica along with specimens from Cuba, Porto Rico, Trinidad, Central America, and South America. However, in their Flora of Jaimaica, Fawcett & Rendle (7: 101. 1936) point out that the localization of Shakespear’s specimen for Jamaica is doubtful and may have been wrongly ascribed to that country by Swartz.
In the present treatment I regard P. involucrata Sw. merely as a superfluous name in the synonymy of P. officinalis, with which species it was identified by Willdenow, but the plants from Cuba referred to by Urban (Symb. 7: 473. 1913) pertain to P. hoffmannseggiana var tribracteata (P. tribracteata) and the reference by him to South American material pertains to both P. barbiflora and P. hoffmannseggiana. Of these two latter species, P. barbiflora is common in Central America and South America, whereas P. hoffmannseggiana (sensu stricto) pertains only to South American plants.
Salzmann’s collection from Bahia, Brazil, upon which Psychotria barbiflora was based, shows broadly lanceolate to oblong-lanceolate bracts, which are slightly, if at all, extended beyond the head of flowers, whereas the bracts of P. hoffmannseggiana are linear or linear-lanceolate and extend beyond the head of flowers. Although the relative length and width of the bracts of these species may intergrade, I am unable to regard them as one taxon.