Discussion

Taxonomic limits of Centaurea have been controversial. The genus has great morphologic diversity, and studies have revealed much cytologic (e.g., N. Garcia-Jacas et al. 1996) and palynologic (e.g., G. Wagenitz 1955) variation as well. During the nineteenth and twentieth centuries, various taxonomists attempted, with limited success, to divide Centaurea into smaller genera or workable infrageneric taxa. The relations of several satellite genera have been controversial as well. Recent molecular phylogenetic studies (A. Susanna et al. 1995; N. Garcia-Jacas et al. 2000, 2001) have begun to clarify relationships within Centaurea and between Centaurea and other genera. These studies make it clear that Centaurea as traditionally defined is polyphyletic, and that generic boundaries should be realigned if monophyletic taxa are to be recognized. Some taxa traditionally included within Centaurea (e.g., the two native North American species, Centaurea americana and C. rothrockii) fall outside the redefined generic boundaries and are here treated in Plectocephalus. Others usually placed into segregate genera (e.g., Cnicus benedictus) are firmly nested within Centaurea. Because the type species of Centaurea (C. centaurium Linnaeus, an African species) falls outside the main lineage of the genus, a proposal has been made to conserve Centaurea with a different type species (W. Greuter et al. 2001), thereby maintaining the nomenclatural stability of most of the numerous species that do fall within the principal Centaurea clade. Although several Centaurea species are widely established as members of the North American flora, and some of these are widely distributed invasive weeds, some of the taxa listed by J. T. Kartesz and C. A. Meacham (1999) are apparently waifs and not permanent members of the flora. These taxa are discussed informally immediately below. Centaurea aspera Linnaeus (rough star thistle) is known from nineteenth-century collections from ballast piles in New York; it does not appear to be established as a member of the North American flora. It can readily be distinguished from the similar C. diluta: the phyllary appendages are divided into palmately radiating clusters of short spines. Centaurea babylonica Linnaeus has been reported from California as a waif (F. Hrusa et al. 2002); apparently it is not established as a permanent member of the flora. It is a tall (to 3 m), yellow-flowered Centaurea with numerous heads clustered in spiciform arrays. The phyllaries are leathery, and appendages are absent or reduced to spines 1 mm or shorter. Centaurea cineraria Linnaeus (dusty miller), an Italian species known from casual garden escapes (California, Maryland, New York), probably is not permanently established as a member of the North American flora. It is a perennial with pinnately or bipinnately divided, densely gray-tomentose leaves, usually solitary, radiant heads somewhat larger than those of C. stoebe, and purple corollas. Centaurea eriophora Linnaeus, reported from California and Colorado (J. T. Kartesz and C. A. Meacham 1999), would key below to C. sulphurea. It differs from C. sulphurea in having densely arachnoid-tomentose involucres. The California report is based on an early twentieth-century collection from Los Angeles County (A. Davidson 2334, UC). It is unlikely that this species is permanently established as a member of the California flora. The Colorado report is apparently erroneous, referenced to a report of Jacea pratensis Lamarck (= Centaurea jacea Miller), a wholly different plant, in W. A. Weber and R. C. Wittmann (1992). A population of Centaurea trichocephala M. Bieberstein ex Willdenow (featherhead or hairy-head knapweed) was found in the late 1970s in a degraded pasture in eastern Washington (B. F. Roché and C. T. Roché 1991). A weed-control program was instituted, and the plants were successfully eradicated. Although it is apparently not established anywhere in North America, C. trichocephala is listed as a noxious weed in Oregon. These plants resemble C. phrygia in having elongate, pectinate-fringed phyllary appendages. In C. trichocephala the linear-filiform, featherlike appendages are much narrower than the phyllary bodies. Plants of the species spread by horizontal roots. According to Roché and Roché, C. trichocephala is apparently self-sterile; the Oregon plants spread clonally and formed no seeds. Although Cnicus has usually been recognized as a distinctive monotypic genus, it has been merged into Centaurea by various authors (e.g., K. Bremer 1994; G. Wagenitz and F. H. Hellwig 1996). Recent molecular systematic studies (N. Garcia-Jacas et al. 2000) provide additional evidence that it is nested within Centaurea.