DISCUSSION
In the protologue of Karatas amazonica, Baker stated that he had seen the species in flower twice: once at Kew, in 1878, and again in 1886 when he identified a specimen for William Bull. At the first opportunity, Baker did not publish the species since Edouard Morren had already given the taxon a name, " Nidularium amazonicum", and had included the plant under this name (but with no description) in his "Catalogue des Bromeliacees cultivees au Jardin botanique de l'Universite de Liege", in 1873. In the same publication, Morren attributed the name Bromelia amazonica to Jean Jules Linden, who had suggested this epithet. Morren's artist-assistants made a beautiful watercolor of the species which is now a part of the Royal Botanic Gardens collection at Kew.
Only in 1886 did Baker decide to describe the species that Linden introduced to cultivation in the mid 1870s. According to the protologue, this species was already known in horticulture by names such as Aechmea amazonica, as well as those mentioned by Morren (1873). Baker himself (1889) declared that the species had been described on the basis of material that flowered at Kew in 1878, and also based on Bull's material from June 1886 (in that order). However, only the first material is compatible with the exsiccata in Kew and this must necessarily be considered the holotype of N.amazonicum. It is composed of one leaf and several flowers. There are no exsiccata that correspond to the second material mentioned by Baker The drawing that is attached to the holotype was produced by Morren based on a specimen deposited at the herbarium of the University of Liege, dated October 1874. The fruiting structure of this specimen has practically no primary bracts. Baker only saw this drawing at Kew in late 1888 (Baker, 1889) and it is not part of the protologue which is dated 1886.
In 1890, Lindman clearly referred to Morren's (1873) catalogue and revalidated the name Nidularium amazonicum as a new combination for Baker's basionym, attributing it to Linden and Morren. Mez (1891) transferred the species to the genus Canistrum and later Smith (1952) included it in Wittrockia. When it was transferrd to Wittrockia, Smith (1952) confirmed the type to be the material that flowered in the Royal Botanic Gardens, Kew, in 1878. However, the probable clonotypes indicated by Smith & Downs (1979) and expanded by Leme & Brito (1993) should be regarded with extreme caution, and were therefore excluded here. Although the stock of specimens of this taxon was probably very small in Europe in the late 19th century, the exsiccata of these alleged clonotypes do not have enough information to positively associate them with the data in the protologue and the holotype material (excluding the sketch by Morren). Furthermore, Baker (1889) stated that Morren had written on the painting of the species that the material (probably that which produced the exsiccata at Liege) had been sent from "Rio" (i.e., Rio de Janeiro) by Binot, which also does not agree with the information contained in the protologue.
More recently, Leme & Brito (1993) considered Wittrockia smithii to be a synonym of W. amazonica because there were no consistent characteristics to differentiate Reitz's species from what was then known as W. amazonica. In fact, it was merely a case of inconsistent terminology used in the morphological description of the respective type specimens, which, in themselves, were not substantially different.
It is interesting to note that the exsiccata that flowered in April 1873 and January 1874 in Liege were identified at the time, probably by Morren, as N. innocentii, thus representing the onset of the conceptual chaos that has reigned until today.
Nidularium amazonicum and N. innocentii, both with wine-colored leaves abaxially, were often mistaken for one another. Foster (1960) tried to solve the problem by indicating the historical motives that led to confusion of the two taxa. However, the state of chaos endured due to the absence of a precise conceptualization of N. amazonicum, which most certainly is why Reitz published the synonym W.smithii
Although N. innocentii and N. amazonicum are both part of the same "white complex" of species, they belong to different subcomplexes. The petals of N. innocentii form a well-developed basal tube with totally white lobes and no appendages. In contrast, N. amazonicum has petals that are shortly connate at base, greenish white or yellowish with white margins and also with well-developed fimbriate appendages.
Nidularium amazonicum shows a clear morphological affinity to N. krisgreeniae. It is distinguished by its acuminate-caudate leaf blades without transverse lines, inflorescence sunken in the leaf rosette, primary bracts suberect-recurved, and sessile flowers. In the wild, N. amazonicum tallows a color-variation pattern. Some populations have totally green leaves while others are purplish wine especially on the abaxial leaf surface, a trait more common in Santa Catarina. The primary bracts vary from reddish brown to red or reddish orange. When exposed to more sunlight, the inflorescence of N. amazonicum is deeply sunken in the leaf rosette, while in shadier spots in the forest it is more protruding, but it never rises above the leaf rosette as in N. krisgreeniae. The corollas of this taxon also vary in color pattern from whitish green to yellowish (yellowish-green) with white margins.
Nidularium amazonicum is concentrated geographically in the states of southern Brazil (Parana, Santa Catarina and, more diffusely, Rio Grande do Sul ). There is only one collection recorded from Sao Paulo. The species lives in the Atlantic slope forest from sea level to ca. 900 m altitude. It grows mainly on the ground and on rocks, but also occurs occasionally as an epiphyte in the lower layer of the forest. is represented in cultivation by at least five distinct clones.
In Parana and Santa Catarina, N. amazonicum is sympatric with N. innocentii and N. procerum, while in Rio Grande do Sul is grows together with N. jonesianum. —SeeLeme 2000a
