58 Phytologia (Aug 2004) 86(2) 



stem. Based on well supported molecular analyses this smallest member 

 of the Cactaceae represents an isolated lineage, and appears to be the 

 only extant transitional form between the basal grade of subfamilies 

 Pereskioideae-Opuntioideae-Maihuenioideae and a strongly supported 

 clade of more derived cacti (Fig. 1.). At present no other member of the 

 Blossfeldioideae has been identified. 



Whereas distinct morphologies separate subfamilies Opuntioidae, 

 Pereskioideae, Maihuenioideae and Blossfeldioideae, the remainder of 

 species are so morphologically diverse that phyletic subdivision of the 

 group has been difficult because of parallel and convergent evolution 

 in vegetative and floral morphology (Buxbaum 1958; Barthlott and 

 Hunt 1993). Classifications have been confusing and unstable, and 

 circumscription of suprageneric taxa continues to be modified to try to 

 meet modern expectations of monophyly (Buxbaum 1974, Gibson and 

 Nobel 1986;I.O.S. 1986,1990; Barthlott 1988, Barthlott and Hunt 1993). 

 In light of recent molecular studies a review of the entire suprageneric 

 classification of the family, at least at the subfamilial level, seems in 

 order. 



The two lineages comprising the clade sister to Blossfeldioideae are 

 quite distinct (see Fig. 1.) and well supported by high bootstrap values 

 and Bayesian probabilities based on the chloroplast DNA studies of 

 Crozier et al. (2004 in prep.). These results show that the two groups are 

 much more distantly related than are the groups of genera within each of 

 them. It is appropriate to recognize these sister clades at equal rank. In 

 so doing the information content of the classification is increased, and the 

 adoption of six subfamilies is not so numerous as to negate its usefulness. 

 Therefore, I am proposing that the proper application of the autonym 

 Cactoideae belongs to the clade of North American globular cacti that 

 includes Mammillaria mammillaris Haw., the conserved type species 

 of the family. This clade corresponds to tribe Cacteae sensu Barthlott 

 and Hunt (1993), though Backeberg (1966) may have been the first to 

 recognize this monophyletic lineage with his subtribe Boreocactinae, 

 a nomen nudum. The morphologically isolated position of this group 

 from other tribes was pointed out specifically by Barthlott (1988) who 



