Abstract
A number of attributes that has proved useful to delimit subfamilial taxa within the Portulacaceae are critically assessed and some little used ones are introduced. The attributes are polarised by an outgroup comparison and a Wagner tree constructed. Some reversals which occur on this tree are considered unacceptable and branches are moved to eliminate this factor, although a longer tree, in terms of character state changes, is produced. A phylogenetic sequence is proposed on this basis and a new classification of the family attempted. On the basis of these results it is suggested that: (1) the genus Calandrinia s.1. be divided into five genera; (2) the genera Talinaria and Anacampseros are closely related; (3) Talinum may not be a natural genus; (4) the family can be conveniently divided into five tribes. A biogeographical analysis indicates that closer floristic relationships exist between western America and Australia and between eastern America and Africa than between eastern and western America or between Australia and Africa.

This publication has 4 references indexed in Scilit: