ABSTRACT
It is generally acknowledged that the genus Narcissus presents great taxonomic problems, and there have been numerous attempts at its classification. Some authors have taken a very wide view of the concept of each species (e.g., Webb, 1980), resulting in as few as 26 recognised species, while some (e.g., Fernandes, 1969b) have taken a very narrow view which results in the recognition of a great many species (upwards of 60), often involving a complex hierarchy of infraspecific taxa. At the generic level, some researchers have taken a narrow view of the delimitation of genera; for example, in the system devised by Haworth (1831), many of the presently accepted subgenera, sections and subsections were recognised as separate genera (e.g., Hermione for the ‘tazetta’ group, Corbularia for the ‘bulbocodium’ group, Ajax for the ‘pseudonarcissus’ group, Ganymedes for the ‘triandrus’ group, etc.). On the other hand, Herbert (1837) took a rather wider view of the genera, and reduced the 16 genera recognised by Haworth to six, while Spach (1846) went a stage further and treated many of Haworth’s genera as sections of the one genus, Narcissus. Many of the plants that had been described as species up to that time were of unknown (or garden) origin, and it was Baker (1875) who attempted to clarify the situation by considering only wild source material for his classification. All the segregate genera recognised by most of the previous authors were included by him in Narcissus, except for Tapeinanthus, which Baker regarded as sufficiently distinct to uphold at generic level; nowadays it, too, is often ‘sunk’ into Narcissus (e.g., Webb, 1980), although Cullen (1986) maintains it on the basis of the rudimentary corona and near-absence of a perianth tube.
