these two morphological characters are not critical for distinguishing them. On the other hand, 
the distribution of these two characters within the Monimiaceae leads to the conclusion that they 
have evolved independently. 
Thus, searching for the most suitable outgroup for the Lauraceae, the best choices to 
consider would be any taxa in the Monimiaceae that have anthers with valvate dehiscence or 
stamens with paired glands at their base. Evidence from comparative studies on the morphology 
of stamens in the Magnoliidae (Endress & Hufford, 1989) points to the fact that valvate 
dehiscence was apparently more common among primitive angiosperms and thus present in 
several primary independent phylogenetic lines. Meanwhile, stamens with paired appendages are 
restricted to some families in the Laurales, and the Chloranthaceae (if the structures are 
homologous). Thus, the limitation to choose an outgroup from the taxa with appendages at the 
base of the stamens would be advisable, namely the subfamilies Hortoniodeae, 
Atherospermatoideae, and the tribes Monimieae and Peumeae in the subfamily Monimioideae 
(sensu Philipson, 1993). Unfortunately, the phylogenetic relationships among these taxa are not 
resolved to a point that allows their use to define confidently ancestral states for characters in the 
Lauraceae. Thus, the outgroup algorithm suggested by Maddison et al. (1984) cannot be used for 
that purpose, except in few characters where the cladograms presented by Qiu et al. (1993) 
support the analysis at the family level. But other methods for character polarization can be 
applied (Watrous & Wheeler, 1981; Stuessy & Crisci, 1984). 
Based on comparative morphology, Takhtajan (1980: 260) suggested the evolution of 
Lauraceae from primitive Monimiaceae; specifically, he mentions the genus Hortonia as the 
taxon that links both families. This genus has been considered, since Money et al.'s (1950) work, 
Z19 
