A detailed study of the above two taxa does not produce any clear differences between them in floral details. Shape and pubescence of bracts and bracteoles, involucre, calyx teeth, squamellae, pubescence of corolla within and without, as well as hypanthium, show the same conformation and similar variations within both taxa. According to the differences set forth by Müller-Argoviensis in his key (Mart. Fl. Bras. 6(5): 369. 1881), P. poeppigiana and P. barcellana possess a calyx three times longer than the ovary as contrasted with P. tomentosa in which the calyx is equal or nearly equal to the length of the ovary. However, an examination of the isotype collection of Psychotria poeppigiana at NY shows a calyx 1.5-1.7 mm long and an ovary (hypanthium) 1.5 mm long, or with proportions similar to those given for P. tomentosa. Actually, such supposed differences ascribed to these taxa by Müller-Argoviensis have been found to be false and cannot be used in differentiating them.
Müller-Argoviensis also notes (Mart. Fl. Bras. 6(5): 370. 1881) that P. poeppigiana differs from P. tomentosa in having the calyx lobes narrower and longer, as well as glabrous or sparsely and long-pilose above, as contrasted with those of P. tomentosa which are stated to have the calyx lobes narrowly triangular-lanceolate with sericeous hairs at first, but quickly deciduous. Moreover, he describes the ovary (hypanthium) of P. poeppigiana as glabrous and that of P. tomentosa as sericeous-pubescent below or on all parts, but eventually glabrous. It may be stated, however, that none of the above supposed differences have been found to be real ones and vary from one collection to another. No reliance can be placed on characters of the shape of calyx lobes, amount of pubescence of the ovary, or the proportionate length of calyx to ovary between these taxa.
The only character found in separating the above taxa has been the type of pubescence found on the midrib and lateral nerves of the lower side of the leaf blades; in P. poeppigiana subsp poeppigiana (P. tomentosa) the hairs are spreading with loosely and widely spreading pubescence of a villous or hirsute type, whereas in P. poeppigiana subsp barcellana (P. barcellana) the hairs on the midrib and lateral nerves beneath are upwardly appressed and strigose. This same type of pubescence is not correlated, however, with the pubescence occurring on the petioles, upper internodes of the stem, or peduncles, for, although in many cases appressed hairs may occur on these parts, too frequently the hairs are found to be spreading. On the specimen of Spruce 1892, cited under both P. poeppigiana and P. barcellana by Müller-Argoviensis, the hairs on the upper part of the stems and on the petioles are spreading. On the other hand, most of the material of P. tomentosa with spreading pubescence on the lower midrib and lateral nerves shows mainly spreading, loose hairs on the stems, petioles, and peduncles. Some intermediate specimens are found, however, in which the hairs on the lower midrib are not sufficiently strigose to be judged as belonging to P. poeppigiana subsp barcellana nor sufficiently widely spreading to merit being identified as P. poeppigiana subsp poeppigiana. For example, specimens that would fall within this intermediate group are Jenman 5444, Cook 6, De la Cruz 2161 and 1569 from British Guiana, which are assigned to P. poeppigiana subsp barcellana, but the hairs of the lower midrib are not appressed to the extent found in typical subsp barcellana. In such cases, the lateral nerves have retained the appressed-strigose condition to indicate the relationship with the subsp barcellana. In the collection of Cook 6, the midrib has hairs somewhat spreading, but the lateral nerves have the hairs more ascending and forwardly projecting as in subsp barcellana. However, in nearly all cases the two subspecies may be readily distinguished by their pubescence on the lower midrib and lateral nerves.
I have assigned subspecific status to these taxa, since they appear to show a definite geographical segregation in various parts of their range. In Central America only the typical P. poeppigiana subsp poeppigiana with spreading hairs occurs, while in Venezuela the spreading-hair type is mainly concentrated in the Andes and Coastal Cordillera, but is replaced in the Guayana of southern Venezuela in the states of Bolívar and Territory Federal Amazonas by the appressed type of pubescence of subsp barcellana. From French Guiana, where the type with spreading hairs only is found, i e P. poeppigiana subsp poeppigiana, the same variation continues in the adjacent lowland areas of British Guiana and Suriname, and in adjacent northeastern Brazil in Territory Amapa and Para state, but in Brazil along the upper Rio Negro above Manaus is replaced by the variation with appressed pubescence, i e subsp barcellana. The type with spreading hairs is common in most of Colombia, Ecuador, Peru, Bolivia, northern Argentina, and adjacent southwestern and central Brazil, penetrating into Mato Grosso and Goias (on the Planalto of Mato Grosso and elsewhere where it often develops narrower leaves proportionately to its length). However, no constancy is found as regards a tendency to a narrow leaf, as witness the contrasting collections of Prance, Silva & Pires 59099 and Irwin, Grear, Souza & dos Santos 16123 from the same area of Serra do Roncador.
Both Bremekamp (Rubiaceae in Pulle, Fl. Surin. 4(1): 255. 1934) and Molina (Ceiba 4(1): 20. 1953) describe the calyx lobes of Cephaelis tomentosa ( = Psychotria poeppigiana subsp poeppigiana) as ciliate, while Bremekamp also adds that they are eglandular. However, dissections of numerous specimens clearly manifest that the calyx lobes are conspicuously glandular with prominent squamellar appendages in the sinuses of the calyx lobes, and, furthermore, that the calyx lobes are glabrous and eciliate throughout. Müller-Argoviensis, on the other hand, correctly describes the calyx lobes of Psychotria tomentosa (= P. poeppigiana subsp poeppigiana) as “basi cum glandulis alternantibus” (Mart. Fl. Bras. 6(5): 370. 1881), but describes those of P. barcellana (— P. poeppigiana subsp barcellana) as “ciliato-piligeris.” It would appear that the erroneous observations by Bremekamp and Molina concerning the calyx lobes as ciliate may be traced to the difficulty of dissection of the corolla and calyx. This is explained by the fact that the base of the corolla tube is closely adherent to the base of the calyx tube and to the top of the hypanthium, making it difficult to separate the delicate hyaline calyx lobes and remainder of the calyx which adhere to the base of the corolla tube. Furthermore, the long, ascending hairs at the base of the hypanthium do not extend up on the calyx tube, but remain at the base of the hypanthium. However, their length may overlap onto the calyx and calyx lobes, thus producing the false impression that they are part of the calyx lobes and calyx tube.