54 Phytologia (Aug 2004) 86(2) 



1 oo 





69 







1 00 



1 .00 





1 00 



1 .00 



1 00 





1 n 





1 .0 











1 00 







.90 









1 



1 



1 .00 







1 .0 









1 .00 















.99 







Caotoidaa* 



Rhipsalidoid*a* 



Bloaffrldioid* a* 



F'-'laih u* nio id * a* 



punt loi da s* 



P*r*=kioid*s* 



Portulaoaoaa* 



Figure 1. Relationships of Cactaceae subfamilies. Congruent topologies 

 resulted from maximum parsimony and Bayesian phylogenetic analyses 

 of 16,620 base pair chloroplast data set comparing 98 cactus species and 

 3 outgroup taxa (simplified from Crozier et al. 2004 in prep.). Bootstrap 

 values shown above branches; Bayesian probabilities shown below. For 

 one of the 52 most parsimonious trees the Consistency Index excluding 

 uninformative characters = 0.48, Retention Index = 0.78. 



However, the molecular study ofNy feller (2002) was unable to support 

 the monophyly of Pereskia. A broader sample of six Pereskia species 

 in the study by Crozier et al. (2004 in prep.) does form a monophyletic 

 clade with moderate bootstrap support and significant Bayesian 

 probability (see Fig. 1). Together the Opuntioideae, Pereskioideae and 

 Maihuenioideae represent less than 15% of the species of the family. 

 The rest of the family, a morphologically diverse group of more than 

 1100 species (Hunt 1999), has traditionally been lumped into the single 

 subfamily Cactoideae based on the absence of synapomorphies defining 

 the Pereskioideae and Opuntioideae. This diversity has usually been 

 subdivided into 8 or 9 tribes (see Barthlott 1988 for a review; Barthlott 

 and Hunt 1993). However, comparative analyses of chloroplast DNA 



