DISCUSSION
According to the original work, the type of Canistrum roseum - the basionym of var. rosea - flowered in December 1879; it came from the Count of Germiny's collection and was presented to E. Morren by Schlumberger.The holotype was examined by means of a photograph provided by J. R. Grant, since the institution where this material is kept (visited by Grant) does not make international loans as was mentioned above.The original description, Morren's painting and the photograph of the type are sufficient to characterize this taxon. It differs from var. lindenii in its leaves which are often reddish, dark red, or with narrow longitudinal dark red stripes, especially in young plants, and also in its pink to red primary and involucral bracts. The variety rosea may also have white or green sepals.
Aechmea fusca, based on sketches by Morren who named it Canistrum fuscum (later made legitimate by Mez), is maintained as a synonym of var. rosea, following the lead of Smith (1955) who qualified this treatment with a question mark. Although Baker (1889) described this taxon's bracts as green, in Morren's painting have a rather brownish hue (fuscus = dark, brown) which is more in agreement with the color pattern of var. rosea
The description of Canistrum binotii Mez, although based on herbarium material, leaves no doubt as to its status as a synonym of E. lindenii var. rosea because of the characteristics of the leaves (60 cm long), the inflorescence (12 cm in diameter), involucral bracts "ex sicco triste purpureis", with laxly arranged, delicate spines, and flowers and floral bracts covered by a dark wool.
The horticultural name Canistrum blockii was also placed on the list of synonyms, based on Chevalier's (1931) statement that this plant was similar to C. roseum.
Finally, for the same reasons that the nomenclatural forms proposed by Reitz (1950, 1952) were made synonymous with the type variety, the forms procerum and humile were also placed in synonymy. As a matter of fact, form procerum is also an illegitimate name because it was based on the type of C. roseum disregarding the autonymy rules (see discussion above).
The geographic distribution of var. lindenii is much narrower than that of var. rosea. It has been reported for Rio de Janeiro, Santa Catarina, Parana and Rio Grande do Sul, and probably occurs also in Sao Paulo since it is a typical Atlantic forest species from moist lowlands and the lower slopes of the coastal mountain range. In Rio Grande do Sul, var. lindenii and var. rosea are sympatric and may be found in the same area (J. C. da Silva, pers. comm.). There are no records of var. lindenii in protected conservation units. —SeeLeme 1997a