Bentham (1880) agreed and strengthened with the observation that this character correlates with 
the presence or absence of extrorse anthers in the third whorl. Pax (1889) thought that the 
number of sporangia in anthers was the most informative character for recognizing major groups 
within the family. Below the first or second division, characters such as anther dehiscence 
(introrse/extrorse), extent of hypanthium development in fruit, extent of development of fourth 
whorl staminodia, and number of sporangia in anthers, were used by these authors in different 
orders of importance to segregate smaller groups. Kostermans (1957) considered that the extent 
of hypanthium development in the fruit was the most meaningful character to understand the 
evolution of Lauraceae. Aware of the variation in number of anther sporangia and changes from 
fertile to sterile stamen whorls in closely related taxa, he underscored these features as some of 
the most important to identify major groups in the family; yet he used them to discern minor 
groups. Hutchinson (1964) argued that in spite of the work done during the first half of this 
century, it was better to follow earlier authors for concepts of taxa below family level; in general, 
his scheme matches Bentham's. Rohwer (1993a), in the most recent contribution dealing with the 
classification of Lauraceae, has attempted the combination of other kinds of information 
(anatomical, phytochemical, embryological) besides morphology in redefining the taxonomical 
units within the family. However, below the subfamily level he only indicated groups of genera 
and did not give them any systematic meaning on the grounds that, in his opinion, reticulate 
variation (of combination of characters) prevents the absolute delimitation of genera and 
suprageneric taxa. 
In order to study the relationships between the group of species studied here and the 
species of Cinnamomum in the Old World, as well as the relationships among the species in the 
216 
